Integral geometry in complex space forms
Integral geometry in complex space forms
Integral geometry in complex space forms
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
<strong>Integral</strong> <strong>geometry</strong> <strong>in</strong> <strong>complex</strong> <strong>space</strong> <strong>forms</strong><br />
Judit Abardia Bochaca<br />
October 2009<br />
Memòria presentada per aspirar al grau de<br />
Doctor en Ciències Matemàtiques.<br />
Departament de Matemàtiques de la Universitat<br />
Autònoma de Barcelona.<br />
Directors:<br />
Eduardo Gallego Gómez i Gil Solanes Farrés
CERTIFIQUEM que la present Memòria ha estat realitzada per na<br />
Judit Abardia Bochaca, sota la direcció dels Drs. Eduardo Gallego<br />
Gómez i Gil Solanes Farrés.<br />
Bellaterra, octubre del 2009<br />
Signat: Dr. Eduardo Gallego Gómez i Dr. Gil Solanes Farrés
Contents<br />
Introduction 1<br />
1 Spaces of constant holomorphic curvature 7<br />
1.1 First def<strong>in</strong>itions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7<br />
1.2 Projective model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9<br />
1.2.1 Po<strong>in</strong>ts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9<br />
1.2.2 Tangent <strong>space</strong> . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10<br />
1.2.3 Metric . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10<br />
1.2.4 Geodesics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11<br />
1.2.5 Isometries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11<br />
1.2.6 Structure of homogeneous <strong>space</strong> . . . . . . . . . . . . . . . . . . . . . . 13<br />
1.3 Mov<strong>in</strong>g frames . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14<br />
1.4 Submanifolds . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17<br />
1.4.1 Totally geodesic submanifolds . . . . . . . . . . . . . . . . . . . . . . . . 17<br />
1.4.2 Geodesic balls . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18<br />
1.5 Space of <strong>complex</strong> r-planes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19<br />
1.5.1 Expression for the <strong>in</strong>variant density <strong>in</strong> terms of a parametrization . . . 20<br />
1.5.2 Density of <strong>complex</strong> r-planes conta<strong>in</strong><strong>in</strong>g a fixed <strong>complex</strong> q-plane . . . . . 25<br />
1.5.3 Density of <strong>complex</strong> q-planes conta<strong>in</strong>ed <strong>in</strong> a fixed <strong>complex</strong> r-plane . . . . 25<br />
1.5.4 Measure of <strong>complex</strong> r-planes <strong>in</strong>tersect<strong>in</strong>g a geodesic ball . . . . . . . . . 25<br />
1.5.5 Reproductive property of Quermass<strong>in</strong>tegrale . . . . . . . . . . . . . . . . 26<br />
2 Introduction to valuations 29<br />
2.1 Def<strong>in</strong>ition and basic properties . . . . . . . . . . . . . . . . . . . . . . . . . . . 29<br />
2.2 Hadwiger Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32<br />
2.3 Alesker Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34<br />
2.4 Valuations on <strong>complex</strong> <strong>space</strong> <strong>forms</strong> . . . . . . . . . . . . . . . . . . . . . . . . . 37<br />
2.4.1 Smooth valuations on manifolds . . . . . . . . . . . . . . . . . . . . . . 37<br />
2.4.2 Hermitian <strong>in</strong>tr<strong>in</strong>sic volumes . . . . . . . . . . . . . . . . . . . . . . . . . 37<br />
2.4.3 Relation between Hermitian <strong>in</strong>tr<strong>in</strong>sic volumes and the valuations given<br />
by Park . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40<br />
2.4.4 Other curvature <strong>in</strong>tegrals . . . . . . . . . . . . . . . . . . . . . . . . . . 40<br />
2.4.5 Relation between the Hermitian <strong>in</strong>tr<strong>in</strong>sic volumes and the second fundamental<br />
form . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41<br />
3 Average of the mean curvature <strong>in</strong>tegral 45<br />
3.1 Previous lemmas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45<br />
3.2 <strong>Integral</strong> of the r-th mean curvature <strong>in</strong>tegral . . . . . . . . . . . . . . . . . . . . 48<br />
3.3 Mean curvature <strong>in</strong>tegral . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52<br />
3.4 Reproductive valuations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56<br />
v
vi<br />
Contents<br />
3.5 Relation with some valuations def<strong>in</strong>ed by Alesker . . . . . . . . . . . . . . . . . 58<br />
3.6 Example: sphere <strong>in</strong> CK 3 (ɛ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59<br />
4 Gauss-Bonnet Theorem and Crofton formulas for <strong>complex</strong> planes 61<br />
4.1 Variation of the Hermitian <strong>in</strong>tr<strong>in</strong>sic volumes . . . . . . . . . . . . . . . . . . . . 61<br />
4.2 Variation of the measure of <strong>complex</strong> r-planes <strong>in</strong>tersect<strong>in</strong>g a doma<strong>in</strong> . . . . . . . 66<br />
4.3 Measure of <strong>complex</strong> r-planes meet<strong>in</strong>g a regular doma<strong>in</strong> . . . . . . . . . . . . . . 68<br />
4.3.1 In the standard Hermitian <strong>space</strong> . . . . . . . . . . . . . . . . . . . . . . 68<br />
4.3.2 In <strong>complex</strong> <strong>space</strong> <strong>forms</strong> . . . . . . . . . . . . . . . . . . . . . . . . . . . 72<br />
4.4 Gauss-Bonnet formula <strong>in</strong> CK n (ɛ) . . . . . . . . . . . . . . . . . . . . . . . . . . 74<br />
4.5 Another method to compute the measure of <strong>complex</strong> l<strong>in</strong>es . . . . . . . . . . . . 76<br />
4.5.1 Measure of <strong>complex</strong> l<strong>in</strong>es meet<strong>in</strong>g a regular doma<strong>in</strong> <strong>in</strong> C n . . . . . . . . 76<br />
4.5.2 Measure of <strong>complex</strong> l<strong>in</strong>es meet<strong>in</strong>g a regular doma<strong>in</strong> <strong>in</strong> CP n and CH n . . 77<br />
4.6 Total Gauss curvature <strong>in</strong>tegral C n . . . . . . . . . . . . . . . . . . . . . . . . . 79<br />
5 Other Crofton formulas 81<br />
5.1 Space of (k, p)-planes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81<br />
5.1.1 Bisectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82<br />
5.2 Variation of the measure of planes meet<strong>in</strong>g a regular doma<strong>in</strong> . . . . . . . . . . 84<br />
5.3 Measure of real geodesics <strong>in</strong> CK n (ɛ) . . . . . . . . . . . . . . . . . . . . . . . . 86<br />
5.4 Measure of real hyperplanes <strong>in</strong> C n . . . . . . . . . . . . . . . . . . . . . . . . . 87<br />
5.5 Measure of coisotropic planes <strong>in</strong> C n . . . . . . . . . . . . . . . . . . . . . . . . . 88<br />
5.6 Measure of Lagrangian planes <strong>in</strong> CK n (ɛ) . . . . . . . . . . . . . . . . . . . . . . 90<br />
Appendix 93<br />
Proof of Theorem 4.3.5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93<br />
Proof of Theorem 4.4.1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100<br />
Bibliography 103<br />
Notation 107<br />
Index 109
Introduction<br />
Classically, <strong>in</strong>tegral <strong>geometry</strong> <strong>in</strong> Euclidean <strong>space</strong> deals with two basic questions: the expression<br />
of the measure of planes meet<strong>in</strong>g a convex doma<strong>in</strong>, the so-called Crofton formulas; and the<br />
study of the measure of movements tak<strong>in</strong>g one convex doma<strong>in</strong> over another fixed convex<br />
doma<strong>in</strong>, the so-called k<strong>in</strong>ematic formula.<br />
In the Euclidean <strong>space</strong> R n , we denote by L r a totally geodesic submanifold of dimension<br />
r, and we call it r-plane. We denote the <strong>space</strong> of r-planes by L r . This <strong>space</strong> has a unique<br />
(up to a constant factor) density <strong>in</strong>variant under the isometry group of R n , denoted by dL r .<br />
Then, given a convex doma<strong>in</strong> Ω ⊂ R n with smooth boundary, the expression of the measure<br />
of r-planes meet<strong>in</strong>g a convex doma<strong>in</strong> is given by<br />
∫<br />
L r<br />
χ(Ω ∩ L r )dL r = c n,r M r−1 (∂Ω), (1)<br />
where c n,r only depends on the dimensions n, r, and M r−1 (∂Ω) denotes the <strong>in</strong>tegral over ∂Ω<br />
of the (r − 1)-th mean curvature <strong>in</strong>tegral.<br />
Thus, the mean curvature <strong>in</strong>tegrals appear naturally <strong>in</strong> the Crofton formula. A classical<br />
known property of mean curvature <strong>in</strong>tegrals is the follow<strong>in</strong>g<br />
∫<br />
M (r)<br />
i<br />
(∂Ω ∩ L r )dL r = c ′ n,r,iM i (∂Ω) (2)<br />
L r<br />
where c ′ (r)<br />
n,r,i only depends on the dimensions n, r, i, and M<br />
i<br />
(∂Ω ∩ L r ) denotes the i-th mean<br />
curvature <strong>in</strong>tegral of ∂Ω ∩ L r as a hypersurface <strong>in</strong> L r<br />
∼ = R r . From (2), it is said that mean<br />
curvature <strong>in</strong>tegrals satisfy a reproductive property.<br />
On the other hand, the k<strong>in</strong>ematic formula <strong>in</strong> R n is expressed as follows. Let Ω 1 and Ω 2 be<br />
two convex doma<strong>in</strong>s with smooth boundary, let O(n) := O(n) ⋉ R n denote the isometry group<br />
of R n , and let dg be an <strong>in</strong>variant density of O(n). Then,<br />
∫<br />
O(n)<br />
χ(Ω 1 ∩ gΩ 2 )dg =<br />
n∑<br />
c n,i M i (∂Ω 1 )M n−i (∂Ω 2 ). (3)<br />
i=0<br />
The previous three formulas were extended to projective and hyperbolic <strong>space</strong>s (cf. [San04]),<br />
i.e. they are known <strong>in</strong> the <strong>space</strong>s of constant sectional curvature k. The generalization of <strong>in</strong>tegral<br />
(2) does not depend on k but <strong>in</strong> the expression (1) for projective and hyperbolic <strong>space</strong><br />
appear other terms, depend<strong>in</strong>g on k. Moreover, its expression depends on the parity of the<br />
dimension of the planes. If r is even, then<br />
∫<br />
L r<br />
χ(Ω ∩ L r )dL r = c n,r−1 M r−1 (∂Ω) + c n,r−3 M r−3 (∂Ω) + · · · + c n,1 M 1 (∂Ω) + c n vol(Ω), (4)<br />
and if r is odd<br />
∫<br />
L r<br />
χ(Ω ∩ L r )dL r = c n,r−1 M r−1 (∂Ω) + c n,r−3 M r−3 (∂Ω) + · · · + c n,2 M 2 (∂Ω) + c n vol(∂Ω), (5)<br />
1
2 Introduction<br />
where c n,j depends on the dimensions n and j and are multiples of k n−j .<br />
The facts that the expression depends on the parity, and that we study an <strong>in</strong>tegral of<br />
the Euler characteristic, rema<strong>in</strong> us the Gauss-Bonnet formula <strong>in</strong> <strong>space</strong>s of constant sectional<br />
curvature, which also depends on the parity of the ambient <strong>space</strong>. We recall here this formula<br />
<strong>in</strong> a <strong>space</strong> of constant sectional curvature k and dimension n.<br />
If n is even, then<br />
M n−1 (∂Ω) + c n−3 M n−3 (∂Ω) + · · · + c 1 M 1 (∂Ω) + k n/2 vol(Ω) = vol(S n−1 )χ(Ω),<br />
and if n is odd,<br />
M n−1 (∂Ω) + c n−3 M n−3 (∂Ω) + · · · + c 2 M 2 (∂Ω) + k (n−1)/2 vol(∂Ω) = vol(Sn−1 )<br />
χ(Ω)<br />
2<br />
where c i depends only on the dimensions n, i and are multiples of the sectional curvature k.<br />
Now, us<strong>in</strong>g the expression (2) and the Gauss-Bonnet formula, we get (4) and (5).<br />
The goal of this work is generalize formulas (1) and (2) <strong>in</strong> the standard Hermitian <strong>space</strong><br />
C n , <strong>in</strong> the <strong>complex</strong> projective <strong>space</strong> and <strong>in</strong> the <strong>complex</strong> hyperbolic <strong>space</strong>, denoted by CK n (ɛ)<br />
with 4ɛ the holomorphic curvature of the manifold (see Section 1.1).<br />
In order to achieve this goal, we use the notion of valuation <strong>in</strong> a vector <strong>space</strong> V , a realvalued<br />
functional φ from the <strong>space</strong> of convex compact doma<strong>in</strong>s K(V ) <strong>in</strong> V to R satisfy<strong>in</strong>g the<br />
follow<strong>in</strong>g additive property<br />
φ(A ∪ B) = φ(A) + φ(B) − φ(A ∩ B)<br />
whenever A, B, A ∪ B ∈ K(V ).<br />
The first examples of valuations are the volume of the convex doma<strong>in</strong>, the area of the<br />
boundary, and the Euler characteristic. Other classical examples of valuations are the socalled<br />
<strong>in</strong>tr<strong>in</strong>sic volumes. They are def<strong>in</strong>ed from the Ste<strong>in</strong>er formula: given a convex doma<strong>in</strong><br />
Ω ⊂ R n , if we denote by Ω r the parallel doma<strong>in</strong> at a distance r, the Ste<strong>in</strong>er formula relates<br />
the volume of Ω r with the so-called <strong>in</strong>tr<strong>in</strong>sic volumes V i (Ω) by<br />
vol(Ω r ) =<br />
n∑<br />
r n−i ω n−i V i (Ω)<br />
i=0<br />
where ω n−i denotes the volume of the (n − i)-dimensional Euclidean ball with radius 1 (cf.<br />
Proposition 2.1.3).<br />
If Ω ⊂ R n is a convex doma<strong>in</strong> with smooth boundary, then <strong>in</strong>tr<strong>in</strong>sic volumes satisfy<br />
V i (Ω) = cM n−i−1 (∂Ω),<br />
and they are the natural generalization of the mean curvature <strong>in</strong>tegrals for non-smooth convex<br />
doma<strong>in</strong>s.<br />
Hadwiger <strong>in</strong> [Had57] proved that all cont<strong>in</strong>uous valuations <strong>in</strong> R n <strong>in</strong>variant under the isometry<br />
group of R n are l<strong>in</strong>ear comb<strong>in</strong>ation of the volume of the convex doma<strong>in</strong>, the area of<br />
the boundary, and the <strong>in</strong>tr<strong>in</strong>sic volumes (see Section 2.2.1). This result has as immediate<br />
consequence formulas (1), (2) and (3).<br />
Alesker <strong>in</strong> [Ale03] proved that the dimension of the <strong>space</strong> of cont<strong>in</strong>uous valuations <strong>in</strong> C n<br />
<strong>in</strong>variant under the holomorphic isometry group of C n is ( )<br />
n+2<br />
2 and gave a basis of this <strong>space</strong>.<br />
In the recent paper of Bernig and Fu, [BF08], there are given other basis of valuations <strong>in</strong> C n .<br />
In particular, the Hermitian <strong>in</strong>tr<strong>in</strong>sic volumes are def<strong>in</strong>ed (see Section 2.4.2). These are the<br />
valuations we will use to work with. The fact that the dimension of the <strong>space</strong> of cont<strong>in</strong>uous<br />
valuations <strong>in</strong>variant under the isometry group of C n is bigger than the one of R 2n is not
3<br />
surpris<strong>in</strong>g if we recall that the holomorphic isometry group of C n , U(n), is smaller than the<br />
isometry group of R 2n , O(2n).<br />
Hermitian <strong>in</strong>tr<strong>in</strong>sic volumes are a k<strong>in</strong>d of generalization of mean curvature <strong>in</strong>tegrals, but<br />
tak<strong>in</strong>g <strong>in</strong>to account that C n has a <strong>complex</strong> structure which def<strong>in</strong>es a canonical vector field on<br />
hypersurfaces. Indeed, at each po<strong>in</strong>t x of a hypersurface, if we consider the normal vector, and<br />
we apply the <strong>complex</strong> structure, then we get a dist<strong>in</strong>guished vector JN <strong>in</strong> the tangent <strong>space</strong><br />
of the hypersurface at x. Moreover, the orthogonal <strong>space</strong> to JN <strong>in</strong> the tangent <strong>space</strong> def<strong>in</strong>es<br />
a <strong>complex</strong> <strong>space</strong> of maximum dimension, n − 1.<br />
So, if S is a smooth hypersurface <strong>in</strong> C n , we can consider the <strong>in</strong>tegral<br />
∫<br />
k n (JN)dx<br />
S<br />
where k n (JN) denotes the normal curvature <strong>in</strong> the direction JN, and this is a valuation <strong>in</strong><br />
C n . Other valuations related to normal curvature of the direction JN appear as elements <strong>in</strong><br />
of Hermitian <strong>in</strong>tr<strong>in</strong>sic volumes basis.<br />
The notion of valuation can be also def<strong>in</strong>ed <strong>in</strong> a differentiable manifold (see Def<strong>in</strong>ition<br />
2.4.1). In real <strong>space</strong> <strong>forms</strong> the volume of a convex doma<strong>in</strong> and the area of its boundary are<br />
valuations. But, it is not known an analogous result to Hadwiger Theorem <strong>in</strong> these <strong>space</strong>s.<br />
The def<strong>in</strong>ition of the Hermitian <strong>in</strong>tr<strong>in</strong>sic volumes can be extended to other <strong>space</strong> of constant<br />
holomorphic curvature. We denote by {µ k,q } the Hermitian <strong>in</strong>tr<strong>in</strong>sic volumes. The subscript<br />
k denotes the degree of the valuation (see Section 2.4.2).<br />
In order to give a similar expression of (1) and (2) <strong>in</strong> the <strong>space</strong>s of constant holomorphic<br />
curvature, we need to describe the <strong>in</strong>tegration <strong>space</strong>. Note that <strong>in</strong> <strong>space</strong>s of constant sectional<br />
curvature we <strong>in</strong>tegrate over the <strong>space</strong> of r-planes, i.e. totally geodesic submanifold of fixed dimension.<br />
In <strong>space</strong>s of constant holomorphic curvature, complete totally geodesic submanifolds<br />
are classified. If ɛ ≠ 0 they are <strong>complex</strong> submanifolds isometric to CK r (ɛ) ⊂ CK n (ɛ), with<br />
1 ≤ r < n or totally real submanifolds isometric to RK q (ɛ) ⊂ CK n (ɛ) with 1 ≤ q ≤ n, where<br />
RK q (ɛ) denotes the <strong>space</strong> of constant sectional curvature ɛ. For ɛ = 0 there are other totally<br />
geodesic submanifolds. We denote the <strong>space</strong> of <strong>complex</strong> planes with <strong>complex</strong> dimension r,<br />
1 ≤ r < n, by L C r , and the <strong>space</strong> of totally real planes of maximum dimension n by L R n, the<br />
so-called Lagrangian manifolds.<br />
In this work, we obta<strong>in</strong> a Crofton formula for <strong>complex</strong> r-planes and Lagrangian planes<br />
∫<br />
( ) n − 1<br />
χ(Ω ∩ L r )dL r = vol(G C n−1,r)<br />
−1· (6)<br />
L C r<br />
· (<br />
n−1<br />
∑<br />
k=n−r<br />
ɛ k−(n−r) ω 2n−2k<br />
( n<br />
k<br />
+ ɛ r (r + 1)vol(Ω)),<br />
) −1<br />
⎛<br />
⎝<br />
r<br />
∑k−1<br />
q=max{0,2k−n}<br />
( 2k−2q<br />
)<br />
⎞<br />
k−q<br />
4 k−q µ 2k,q (Ω) + (k + r − n + 1)µ 2k,k (Ω) ⎠<br />
and<br />
∫<br />
L R n<br />
⎛<br />
· ⎝<br />
∫<br />
L R n<br />
χ(Ω ∩ L)dL = vol(G 2n,n)ω n<br />
n!<br />
n−1<br />
2∑<br />
q=0<br />
( ) 2q − 1 −1<br />
4 q−n<br />
q − 1 2q + 1 µ n,q(Ω) if n is odd, (7)<br />
χ(Ω ∩ L)dL = vol(G 2n,n)<br />
· (8)<br />
n!<br />
n<br />
2∑<br />
( ) 2q − 1 −1<br />
4 q−n n<br />
⎞<br />
ω n<br />
2∑<br />
( ) n −1<br />
q − 1 2q + 1 µ n,q(Ω) + ɛ i 2 −n+1 ω n−2i<br />
n<br />
2 + i µ<br />
n + 1 n+2i,<br />
n<br />
2 +i (Ω) ⎠ if n is even,<br />
q=0<br />
i=1
4 Introduction<br />
where ω i denotes the volume of the i-dimensional Euclidean unit ball.<br />
Previous formulas have more addends that the correspond<strong>in</strong>g ones <strong>in</strong> the <strong>space</strong>s of constant<br />
sectional curvature, but they are similar. If ɛ = 0, i.e. <strong>in</strong> C n also appear all the valuations<br />
with the correspond<strong>in</strong>g degree. If ɛ ≠ 0 the notion of degree of a valuation has no sense but<br />
there is a similitude with the expression <strong>in</strong> <strong>space</strong>s of constant sectional curvature compar<strong>in</strong>g<br />
the subscripts of the valuations.<br />
In order to get these expressions we use a variational method. That is, we take a smooth<br />
vector field X def<strong>in</strong>ed on the manifold and we consider its flow φ t . We prove the follow<strong>in</strong>g<br />
formula of first variation<br />
∫<br />
∫ ∫<br />
d<br />
dt∣ χ(φ t (Ω) ∩ L r )dL r = 〈X, N〉<br />
σ 2r (II| V )dV dp<br />
t=0 G C n−1,r (Dp)<br />
L C r<br />
∂Ω<br />
where N is the exterior normal field, D is the distribution <strong>in</strong> the tangent <strong>space</strong> at ∂Ω orthogonal<br />
to JN, and σ 2r (II| V ) denotes the 2r-th symmetric elementary function of the second<br />
fundamental form II restricted to V ∈ G C n−1,r (D p), the Grassmannian of <strong>complex</strong> planes with<br />
<strong>complex</strong> dimension r <strong>in</strong>side D p .<br />
On the other hand, we get an expression of the variation of valuations µ k,q us<strong>in</strong>g the method<br />
<strong>in</strong> [BF08].<br />
Compar<strong>in</strong>g both variations and solv<strong>in</strong>g a system of l<strong>in</strong>ear equations we obta<strong>in</strong> the result.<br />
Us<strong>in</strong>g the same variational method we also obta<strong>in</strong> a Gauss-Bonnet formula for the <strong>space</strong>s<br />
of constant holomorphic curvature. It is known that the variation of the Euler characteristic<br />
is zero. Thus, we can express it as a sum of Hermitian <strong>in</strong>tr<strong>in</strong>sic volumes such that its variation<br />
vanishes. The obta<strong>in</strong>ed Gauss-Bonnet formula is the follow<strong>in</strong>g<br />
⎛<br />
n−1<br />
∑<br />
ω 2n χ(Ω) = (n + 1)ɛ n ɛ c ω 2n−2c (n − c)<br />
vol(Ω) +<br />
n ( ) ⎝<br />
n−1<br />
c<br />
c=0<br />
∑c−1<br />
q=max{0,2c−n}<br />
( 2c−2q<br />
)<br />
⎞<br />
c−q<br />
4 c−q µ 2c,q + (c + 1)µ 2c,c<br />
⎠.<br />
In <strong>space</strong>s of constant sectional curvature k, Solanes <strong>in</strong> [Sol06] related the measure of planes<br />
meet<strong>in</strong>g a doma<strong>in</strong> with the Euler characteristic of the doma<strong>in</strong><br />
In CK n (ɛ), we get<br />
ω n χ(Ω) = 1 n M n−1(∂Ω) +<br />
2k ∫<br />
nω n−1<br />
L n−2<br />
χ(Ω ∩ L n−2 )dL n−2 .<br />
ω 2n χ(Ω) = 1<br />
∫<br />
2n M 2n−1(∂Ω) + ɛ χ(Ω ∩ L n−1 )dL n−1 +<br />
L C n−1<br />
n∑<br />
j=1<br />
ɛ j ω 2n<br />
µ 2j,j (Ω).<br />
ω 2j<br />
(9)<br />
The analogous expression to (2), it is given when we <strong>in</strong>tegrate the mean curvature <strong>in</strong>tegral<br />
over <strong>complex</strong> r-planes. The obta<strong>in</strong>ed expression for a compact oriented (possible with<br />
boundary) hypersurface S of class C 2 is<br />
∫<br />
L C r<br />
M (r)<br />
1 (S∩L r)dL r = ω 2n−2vol(G C n−2,r−1 ) ( n −1 (<br />
(2n − 1)<br />
2r(2r − 1) r) 2nr − n − r ∫<br />
M 1 (S) +<br />
n − r<br />
where k n (JN) denotes the normal curvature <strong>in</strong> the direction JN ∈ T S.<br />
S<br />
)<br />
k n (JN)<br />
(10)
5<br />
To get this result, first we obta<strong>in</strong>, us<strong>in</strong>g mov<strong>in</strong>g frames, the follow<strong>in</strong>g <strong>in</strong>termediate expression<br />
(for any mean curvature <strong>in</strong>tegral). If r, i ∈ N such that 1 ≤ r ≤ n and 0 ≤ i ≤ 2r − 1,<br />
then<br />
∫<br />
M (r)<br />
i<br />
(S ∩ L r )dL r (11)<br />
L C r<br />
( ) 2r − 1 −1 ∫<br />
=<br />
i<br />
S<br />
∫RP 2n−2 ∫<br />
G C n−2,r−1<br />
|〈JN, e r 〉| 2r−i<br />
(1 − 〈JN, e r 〉 2 ) r−1 σ i(p; e r ⊕ V )dV de r dp,<br />
where e r ∈ T p S unit vector, V denotes a <strong>complex</strong> (r − 1)-plane conta<strong>in</strong><strong>in</strong>g p and conta<strong>in</strong>ed <strong>in</strong><br />
{N, JN, e r , Je r } ⊥ , σ i (p; e r ⊕ V ) denotes the i-th symmetric elementary function of the second<br />
fundamental form of S restricted to the real sub<strong>space</strong> e r ⊕ V and the <strong>in</strong>tegration over RP 2n−2<br />
denotes the projective <strong>space</strong> of the unit tangent <strong>space</strong> of the hypersurface.<br />
In order to complete the generalization of equation (1) <strong>in</strong> CK n (ɛ), it rema<strong>in</strong>s to study the<br />
measure of (non-maximal) totally real planes. These are the other totally geodesic submanifolds<br />
of CK n (ɛ), ɛ ≠ 0. Us<strong>in</strong>g the same techniques as <strong>in</strong> the rest of this work, it does not<br />
seem possible to solve this case s<strong>in</strong>ce we cannot obta<strong>in</strong> enough <strong>in</strong>formation of the variational<br />
properties of the measures of totally real planes meet<strong>in</strong>g a doma<strong>in</strong> <strong>in</strong> CK n (ɛ).<br />
On the other hand, it would be <strong>in</strong>terest<strong>in</strong>g to extend formula (10) to i ∈ {2, . . . , 2r − 1}.<br />
Next we expla<strong>in</strong> the organization of the text.<br />
Chapter 1 conta<strong>in</strong>s a description of the <strong>space</strong>s of constant holomorphic curvature. We<br />
review its def<strong>in</strong>ition and describe some of the most important submanifolds, i.e. the totally<br />
geodesic submanifolds, the geodesic spheres, and the <strong>complex</strong> planes. In this chapter we also<br />
recall the method of mov<strong>in</strong>g frames, which will be used along this text. Us<strong>in</strong>g mov<strong>in</strong>g frames,<br />
we give an expression for the density of the <strong>space</strong> of <strong>complex</strong> planes. F<strong>in</strong>ally, we prove that<br />
<strong>in</strong>tegral ∫ L χ(Ω ∩ L r)dL C r satisfies a reproductive property.<br />
r<br />
Chapter 2 is devoted to the study of valuations <strong>in</strong> C n and <strong>in</strong> the <strong>space</strong>s of constant holomorphic<br />
curvature. First of all, we review the concept and the ma<strong>in</strong> properties of the valuations<br />
on R n together with the Hadwiger Theorem (which characterize all cont<strong>in</strong>uous valuations <strong>in</strong><br />
R n <strong>in</strong>variants under the isometry group). An analogous Hadwiger Theorem <strong>in</strong> C n is stated.<br />
F<strong>in</strong>ally, we def<strong>in</strong>e the used valuations <strong>in</strong> this work <strong>in</strong> <strong>space</strong>s of constant holomorphic curvature,<br />
and we give new properties and relations with other valuations also important <strong>in</strong> the<br />
next chapters.<br />
Chapter 3 gives a proof of (10). First of all, we prove some geometric lemmas and we<br />
obta<strong>in</strong> the expression for the mean curvature <strong>in</strong>tegrals over the <strong>space</strong> of <strong>complex</strong> planes <strong>in</strong><br />
terms of an <strong>in</strong>tegral over the boundary of the doma<strong>in</strong> given <strong>in</strong> (11). This expression will be<br />
fundamental to atta<strong>in</strong> the goal of this chapter. As a corollary of (10) we characterize the<br />
valuations of degree 2n − 2 satisfy<strong>in</strong>g a reproductive property <strong>in</strong> C n , and we give the relation<br />
among different valuations def<strong>in</strong>ed by Alesker (already reviewed <strong>in</strong> Chapter 2). The results of<br />
this chapter are conta<strong>in</strong>ed <strong>in</strong> [Aba].<br />
In Chapter 4 we obta<strong>in</strong> the measure of <strong>complex</strong> planes <strong>in</strong>tersect<strong>in</strong>g a doma<strong>in</strong> <strong>in</strong> the <strong>space</strong>s of<br />
constant holomorphic curvature <strong>in</strong> terms of the Hermitian <strong>in</strong>tr<strong>in</strong>sic volumes def<strong>in</strong>ed at Chapter<br />
2. We also give an expression of the Gauss-Bonnet formula <strong>in</strong> terms of these valuations. In<br />
order to get these expressions we use a variational method. First, we obta<strong>in</strong> an expression for<br />
the variation of the measure of <strong>complex</strong> planes and for the Hermitian <strong>in</strong>tr<strong>in</strong>sic volumes. In<br />
this chapter, we verify the certa<strong>in</strong>ty of (6). A constructive proof, where we f<strong>in</strong>d the constants<br />
<strong>in</strong> the expression is given <strong>in</strong> the appendix. As a corollary, we express the total mean Gauss<br />
curvature <strong>in</strong> C n also <strong>in</strong> terms of the Hermitic <strong>in</strong>tr<strong>in</strong>sic volumes. F<strong>in</strong>ally, we relate Chapters 3
6 Introduction<br />
and 4 obta<strong>in</strong><strong>in</strong>g another method to compute the measure of <strong>complex</strong> l<strong>in</strong>es meet<strong>in</strong>g a doma<strong>in</strong>.<br />
The results of this chapter are conta<strong>in</strong>ed <strong>in</strong> [AGS09].<br />
Chapter 5 studies the measure of another type of planes meet<strong>in</strong>g a doma<strong>in</strong> <strong>in</strong> C n , the socalled<br />
coisotropic planes. These planes are the orthogonal direct sum of a <strong>complex</strong> sub<strong>space</strong><br />
of <strong>complex</strong> dimension n − p and a totally real sub<strong>space</strong> of dimension p. Totally real planes of<br />
maximum dimension and real hyperplanes are particular cases of this type of planes. Us<strong>in</strong>g<br />
similar techniques as <strong>in</strong> Chapter 4 we give an expression for the measure of planes of this type<br />
meet<strong>in</strong>g a doma<strong>in</strong>. For the <strong>space</strong>s of constant holomorphic curvature we prove (7) and (8),<br />
which give the measure of totally real planes of maximum dimension, the so-called Lagrangian<br />
planes.<br />
The appendix conta<strong>in</strong>s the constructive proof of (6) and (9). That is, we give the method<br />
that allowed us to obta<strong>in</strong> the constants appear<strong>in</strong>g <strong>in</strong> these expressions. This proof consists,<br />
at a f<strong>in</strong>al <strong>in</strong>stance, to solve a l<strong>in</strong>ear system obta<strong>in</strong>ed from the study of the variation of both<br />
sides of the expressions, as it is detailed <strong>in</strong> Chapter 4.<br />
Acknowledgments<br />
In these f<strong>in</strong>al l<strong>in</strong>es, I would like to express my gratitude to all people who, <strong>in</strong> some way, allowed<br />
me to convert this work <strong>in</strong>to a Thesis. Specially, I would like to mention my first director,<br />
Eduardo Gallego, my codirector, Gil Solanes, for their patient and support; and the Group<br />
of Geometry and Topology of the Universitat Autònoma de Barcelona, <strong>in</strong> particular Agustí<br />
Reventós, to <strong>in</strong>troduce me to the “<strong>geometry</strong> world”. I would also like to thank Andreas Bernig<br />
for all his advices, and the members of the Department of Mathematics <strong>in</strong> the University of<br />
Fribourg to offer me such nice work ambience dur<strong>in</strong>g my stay. Although I am not go<strong>in</strong>g to<br />
write down a list with the name of all mathematicians who contributed with their enlighten<strong>in</strong>g<br />
conversations to the completion of this work, and hopefully, to some future work, I would be<br />
very satisfied if they feel identified <strong>in</strong> these l<strong>in</strong>es.<br />
Outside from the direct relation to this work, I would like to thank my colleagues of<br />
doctorate, specially my colleagues <strong>in</strong> the room, and also <strong>in</strong> the <strong>geometry</strong> and topology area,<br />
<strong>in</strong> Barcelona, but also <strong>in</strong> Fribourg. In the same way I would like to thank all the persons I<br />
have met dur<strong>in</strong>g these years and become my friends.<br />
F<strong>in</strong>ally, I thank my parent’s support, not only <strong>in</strong> these years of doctorate.<br />
This work has been written under the support of the “Departament d’Universitats, Recerca<br />
i Societat de la Informació de la Generalitat de Catalunya”, the European Social Found and<br />
the Swiss National Found.
Chapter 1<br />
Spaces of constant holomorphic<br />
curvature<br />
1.1 First def<strong>in</strong>itions<br />
In this section we <strong>in</strong>troduce the <strong>space</strong>s of constant holomorphic curvature, also called <strong>complex</strong><br />
<strong>space</strong> <strong>forms</strong>, and give the properties we shall use along this work. First of all, we recall some<br />
basic def<strong>in</strong>itions.<br />
Def<strong>in</strong>ition 1.1.1. Let M be a differentiable manifold. M is a <strong>complex</strong> manifold if it has an<br />
atlas such that the change of coord<strong>in</strong>ates are holomorphic, that is, {(U α , φ α )} is an atlas with<br />
{U α } an open cover<strong>in</strong>g of M and φ α : U α → C n homeomorphisms such that φ β ◦ φ −1<br />
α are<br />
holomorphic <strong>in</strong> its doma<strong>in</strong> of def<strong>in</strong>ition.<br />
Examples<br />
(i) The vector <strong>space</strong> C n is a <strong>complex</strong> manifold of <strong>complex</strong> dimension n.<br />
(ii) The <strong>complex</strong> projective <strong>space</strong> CP n is a <strong>complex</strong> manifold of <strong>complex</strong> dimension n.<br />
The <strong>complex</strong> projective <strong>space</strong> can be def<strong>in</strong>ed analogously to the real projective <strong>space</strong>.<br />
Let us consider <strong>in</strong> C n+1 \{0} the equivalence relation which identifies the po<strong>in</strong>ts differ<strong>in</strong>g<br />
by a <strong>complex</strong> multiple. Then, we take as an atlas the open sets {U 0 , ..., U n } such that<br />
U j = {(z 0 , ..., z n ) ∈ C n+1 | z j ≠ 0}<br />
and for every U j we take the map φ j (z 0 , ..., z n ) = (z 0 /z j , ..., z j−1 /z j , z j+1 /z j , ..., z n /z j )<br />
which is a homeomorphism. It can be proved that the change of coord<strong>in</strong>ates are holomorphics.<br />
Def<strong>in</strong>ition 1.1.2. Let M be a <strong>complex</strong> manifold. A l<strong>in</strong>ear map J : T x M → T x M is an almost<br />
<strong>complex</strong> structure of M if for each x ∈ M, the restriction of J at T x M satisfies J x : T x M →<br />
T x M, J 2 = −Id, and J varies differentially on M.<br />
Note that any <strong>complex</strong> manifold admits an almost <strong>complex</strong> structure. Indeed, the tangent<br />
<strong>space</strong> of a <strong>complex</strong> manifold has a <strong>complex</strong> vector <strong>space</strong> structure, so the map “multiply by i”<br />
is well-def<strong>in</strong>ed and satisfies that applied twice is the map −Id. We call this canonical almost<br />
<strong>complex</strong> structure <strong>complex</strong> structure.<br />
Def<strong>in</strong>ition 1.1.3. Let V be a <strong>complex</strong> vector <strong>space</strong> and let u, v ∈ V . It is said that h :<br />
V × V → C is an Hermitian product on V if<br />
7
8 Spaces of constant holomorphic curvature<br />
1. it is C-l<strong>in</strong>ear with respect to the first component,<br />
2. h(u, v) = h(u, v).<br />
Remark 1.1.4. From the properties of a Hermitian product, it follows that if λ ∈ C then<br />
h(u, v) = λh(u, v). Indeed, by def<strong>in</strong>ition we get the follow<strong>in</strong>g equalities<br />
h(u, λv) = h(λv, u) = λh(v, u) = λh(u, v).<br />
Def<strong>in</strong>ition 1.1.5. Let M be a differentiable manifold with <strong>complex</strong> structure J and a Riemannian<br />
metric g. Then, g is called a Hermitian metric if it is compatible with the <strong>complex</strong><br />
structure, i.e. it satisfies g x (Ju, Jv) = g x (u, v) for every x ∈ M and u, v ∈ T x M.<br />
Def<strong>in</strong>ition 1.1.6. Let M be a <strong>complex</strong> manifold with <strong>complex</strong> structure J and Hermitian<br />
metric g. The 2-form ω def<strong>in</strong>ed by<br />
is the Kähler form.<br />
ω(u, v) = g(u, Jv), ∀ u, v ∈ T x M<br />
Remark 1.1.7. Given a <strong>complex</strong> manifold M with <strong>complex</strong> structure J and a Hermitian product<br />
def<strong>in</strong>ed on T M, we get a Hermitian metric on M from the real part of the Hermitian product,<br />
and a Kähler form on M from the imag<strong>in</strong>ary part of the Hermitian product.<br />
Def<strong>in</strong>ition 1.1.8. A <strong>complex</strong> manifold M is called a Kähler manifold if it has a Hermitian<br />
metric such that the Kähler form associated to this metric is closed.<br />
Proposition 1.1.9 ([O’N83] page 326). Let M be a Kähler manifold with connection ∇.<br />
Then,<br />
∇JX = J∇X, ∀X ∈ X(M).<br />
Def<strong>in</strong>ition 1.1.10. A sub<strong>space</strong> W ⊂ T x M of <strong>complex</strong> dimension 1 is a <strong>complex</strong> direction or<br />
a holomorphic section of the tangent <strong>space</strong> if W is <strong>in</strong>variant under J, i.e. JW = W .<br />
If w ≠ 0 ∈ W , then the vectors {w, Jw} constitute a basis of W , as a real sub<strong>space</strong>.<br />
Def<strong>in</strong>ition 1.1.11. The holomorphic curvature is the sectional curvature of holomorphic sections.<br />
Def<strong>in</strong>ition 1.1.12. A <strong>space</strong> of constant holomorphic curvature 4ɛ of dimension n is a complete,<br />
simply connected Kähler manifold of <strong>complex</strong> dimension n, such that the holomorphic<br />
curvature is constant and equal to 4ɛ for every po<strong>in</strong>t and every <strong>complex</strong> direction.<br />
Theorem 1.1.13 ([KN69] Theorem 7.9 page 170). Two complete, simply connected Kähler<br />
manifolds with constant holomorphic curvature equal to 4ɛ are holomorphically isometric.<br />
Def<strong>in</strong>ition 1.1.14. We denote by CK n (ɛ) any <strong>space</strong> of constant holomorphic curvature of<br />
dimension n. If ɛ > 0, then it corresponds to the <strong>complex</strong> projective <strong>space</strong> CP n , if ɛ < 0, to<br />
the <strong>complex</strong> hyperbolic <strong>space</strong> CH n , and if ɛ = 0, to the Hermitian standard <strong>space</strong> C n .<br />
Spaces of constant holomorphic curvature are also called <strong>complex</strong> <strong>space</strong> <strong>forms</strong>.<br />
Remark 1.1.15. Complex <strong>space</strong> <strong>forms</strong> are, <strong>in</strong> some sense, a generalization of real <strong>space</strong> <strong>forms</strong><br />
(<strong>space</strong>s of constant sectional curvature), i.e. the complete simply connected Riemannian manifolds<br />
with constant sectional curvature. Real <strong>space</strong> <strong>forms</strong> are (up to isometry) the Euclidean<br />
<strong>space</strong> R n , the real projective <strong>space</strong> RP n and the real hyperbolic <strong>space</strong> H n . The results <strong>in</strong> this<br />
work extend some of the classical results <strong>in</strong> <strong>in</strong>tegral <strong>geometry</strong> from real <strong>space</strong> <strong>forms</strong> to <strong>complex</strong><br />
<strong>space</strong> <strong>forms</strong>. Santaló [San52] and Griffiths [Gri78], among others, obta<strong>in</strong>ed some results<br />
of classical <strong>in</strong>tegral <strong>geometry</strong> <strong>in</strong> the standard Hermitian <strong>space</strong> and <strong>in</strong> the <strong>complex</strong> projective<br />
<strong>space</strong> tak<strong>in</strong>g <strong>complex</strong> submanifolds. In this work, we deal with non-empty doma<strong>in</strong>s and, thus,<br />
with real hypersurfaces.
1.2 Projective model 9<br />
Def<strong>in</strong>ition 1.1.16. Given two planes Π and Π ′ of real dimension 2 <strong>in</strong> a vector <strong>space</strong> with a<br />
scalar product, the angle between the two planes is def<strong>in</strong>ed as the <strong>in</strong>fimum among the angles<br />
between a pair of vectors, one <strong>in</strong> Π and the other one <strong>in</strong> Π ′ .<br />
Def<strong>in</strong>ition 1.1.17. Let Π be a plane with real dimension 2 <strong>in</strong> the tangent <strong>space</strong> of a po<strong>in</strong>t <strong>in</strong> a<br />
Kähler manifold with <strong>complex</strong> structure J. The holomorphic angle µ(Π) is the angle between<br />
Π and J(Π).<br />
Proposition 1.1.18 ([KN69] page 167). Let M be a Kähler manifold with <strong>complex</strong> structure<br />
J and Hermitian metric g. The holomorphic angle of a plane Π ⊂ T x M, x ∈ M, is given by<br />
where u, v form an orthonormal basis of Π.<br />
cos µ(Π) = |g(u, Jv)|<br />
Remark 1.1.19. The holomorphic angle of a plane takes values between 0 and π/2. In the<br />
extreme cases we have holomorphic planes, when the holomorphic angle is 0; and totally real<br />
planes def<strong>in</strong>ed as the planes with holomorphic angle π/2.<br />
In a <strong>complex</strong> <strong>space</strong> form, the sectional curvature of any plane can be computed from the<br />
holomorphic curvature and the holomorphic angle of the plane.<br />
Proposition 1.1.20 ([KN69] page 167). Let M be a Kähler manifold with constant holomorphic<br />
curvature 4ɛ. Then, the sectional curvature of any plane Π ⊂ T x M, x ∈ M is given<br />
by<br />
where µ(Π) is the holomorphic angle of the plane Π.<br />
K(Π) = ɛ ( 1 + 3 cos 2 µ(Π) ) (1.1)<br />
Corollary 1.1.21. Sectional curvature of any plane <strong>in</strong> the tangent <strong>space</strong> of a po<strong>in</strong>t <strong>in</strong> a<br />
<strong>complex</strong> <strong>space</strong> form with constant holomorphic curvature 4ɛ lies <strong>in</strong> the <strong>in</strong>terval [ɛ, 4ɛ], if ɛ > 0<br />
and <strong>in</strong> the <strong>in</strong>terval [4ɛ, ɛ], if ɛ < 0.<br />
1.2 Projective model<br />
Along this work, we shall use the projective model of CK n (ɛ), which we describe here briefly<br />
(cf. [Gol99]).<br />
If ɛ = 0, we are consider<strong>in</strong>g the standard Hermitian <strong>space</strong> C n with the standard Hermitian<br />
product. Along this section we suppose ɛ ≠ 0, unless otherwise stated.<br />
1.2.1 Po<strong>in</strong>ts<br />
Endow C n+1 with the Hermitian product<br />
Def<strong>in</strong>e<br />
(z, w) = sign(ɛ)z 0 w 0 +<br />
n∑<br />
z j w j . (1.2)<br />
j=1<br />
H := {z ∈ C n+1 | (z, z) = ɛ}.<br />
H is a real hypersurface of C n+1 (i.e. it has real dimension 2n + 1). We def<strong>in</strong>e the po<strong>in</strong>ts of<br />
CK n (ɛ) as<br />
CK n (ɛ) := π(H)<br />
where<br />
π : C n+1 \{0} → C n+1 \{0}/C ∗ = CP n . (1.3)
10 Spaces of constant holomorphic curvature<br />
Remarks 1.2.1. (i) The fiber of Π for the po<strong>in</strong>ts <strong>in</strong> π(H) =: CK n (ɛ) is S 1 . Indeed, let z,<br />
w ∈ H such that π(z) = π(w). By def<strong>in</strong>ition of π, we have w = αz. On the other hand,<br />
it holds (z, z) = (αz, αz) = ɛ. Thus, α = e iθ with θ ∈ R and π −1 ([z]) ∼ = S 1 .<br />
(ii) If ɛ > 0, then CK n (ɛ) co<strong>in</strong>cides as a subset with CP n . But, if ɛ < 0, then CK n (ɛ) is an<br />
open set of CP n .<br />
The differentiable structure and the structure of a <strong>complex</strong> manifold we take <strong>in</strong> CK n (ɛ) is<br />
the same as the one of an open set of CP n .<br />
1.2.2 Tangent <strong>space</strong><br />
The tangent <strong>space</strong> of a po<strong>in</strong>t z ∈ H is<br />
T z H = {w ∈ C n+1 | Re(z, w) = 0}.<br />
The elements <strong>in</strong> the tangent <strong>space</strong> of π(z) ∈ CK n (ɛ) are obta<strong>in</strong>ed from the image of the<br />
elements <strong>in</strong> the tangent <strong>space</strong> of the po<strong>in</strong>t z under dπ. Moreover, the kernel of dπ has dimension<br />
1.<br />
The direction that its image under dπ is the null vector is Jz, s<strong>in</strong>ce it is the tangent<br />
direction to the fiber. (Note that Jz ∈ T z H s<strong>in</strong>ce Re(z, Jz) = 0.) Indeed, the fiber of a po<strong>in</strong>t<br />
[z] is {e iθ z | θ ∈ R}, then<br />
∂(e iθ z)<br />
∂θ ∣ = iz = Jz.<br />
θ=0<br />
Thus, the tangent <strong>space</strong> at z ∈ H can be decomposed as<br />
T z H = 〈Jz〉 ⊕ 〈Jz〉 ⊥ .<br />
The tangent <strong>space</strong> at po<strong>in</strong>ts <strong>in</strong> CK n (ɛ) co<strong>in</strong>cides with the image by the differential map of<br />
the projection of vectors 〈Jz〉 ⊥ at T z H.<br />
Given a vector v ∈ T π(z) CK n (ɛ) there are <strong>in</strong>f<strong>in</strong>itely many vectors of T z H such that under<br />
the differential map dπ give the same vector v, but we can dist<strong>in</strong>guish the one ly<strong>in</strong>g <strong>in</strong> 〈Jz〉 ⊥ ,<br />
which is called horizontal lift and we denote by v L . All other vectors are obta<strong>in</strong>ed as l<strong>in</strong>ear<br />
comb<strong>in</strong>ation of this vector and a multiple of Jz.<br />
1.2.3 Metric<br />
Let v, w ∈ T π(z) CH n . The Hermitian product at CK n (ɛ) between v, w is def<strong>in</strong>ed by<br />
(v, w) ɛ := (v L , w L ), (1.4)<br />
that is, the Hermitian product def<strong>in</strong>ed at C n+1 applied to the horizontal lift of the vectors.<br />
The real part of this product gives a Hermitian metric on CK n (ɛ)<br />
〈v, w〉 ɛ := Re(v L , w L ).<br />
This metric co<strong>in</strong>cides with the so-called Fub<strong>in</strong>i-Study metric, if ɛ > 0 and with the so-called<br />
Bergmann metric, if ɛ < 0 (cf. [Gol99, page 74]).<br />
Along this work we denote the Hermitian metric of CK n (ɛ) by 〈 , 〉 <strong>in</strong>stead of 〈 , 〉 ɛ .<br />
Notation 1.2.2. In order to unify the study of the <strong>complex</strong> <strong>space</strong> <strong>forms</strong>, we def<strong>in</strong>e, <strong>in</strong> the<br />
same way as it is classically done <strong>in</strong> real <strong>space</strong> <strong>forms</strong>, the follow<strong>in</strong>g trigonometric generalized<br />
functions<br />
⎧<br />
s<strong>in</strong>(α √ ɛ)<br />
√ , if ɛ > 0<br />
⎪⎨ ɛ<br />
s<strong>in</strong> ɛ (α) = α, if ɛ = 0<br />
s<strong>in</strong>h(α √ −ɛ)<br />
⎪⎩ √ , if ɛ < 0<br />
−ɛ
1.2 Projective model 11<br />
and<br />
⎧<br />
⎨ cos(α √ ɛ), if ɛ > 0<br />
cos ɛ (α) = 1, if ɛ = 0<br />
⎩<br />
cosh(α √ |ɛ|), if ɛ < 0<br />
cot ɛ (α) = cos ɛ(α)<br />
s<strong>in</strong> ɛ (α) .<br />
1.2.4 Geodesics<br />
Geodesics <strong>in</strong> the projective model of CK n (ɛ) are given by the projection of the <strong>in</strong>tersection<br />
po<strong>in</strong>ts between H and a plane <strong>in</strong> C n+1 such that it is spanned by a vector correspond<strong>in</strong>g to a<br />
representative <strong>in</strong> H ⊂ C n+1 of a po<strong>in</strong>t z <strong>in</strong> the geodesic at CK n (ɛ), and a vector u tangent to<br />
the geodesic at z.<br />
Then, the expression of a geodesic at CK n (ɛ) is given by [γ(t)] = [cos ɛ (t)z + s<strong>in</strong> ɛ (t)u] where<br />
u ∈ 〈Jz〉 ⊥ ⊂ T z H.<br />
The distance between two po<strong>in</strong>ts <strong>in</strong> the <strong>complex</strong> projective and hyperbolic <strong>space</strong> can be<br />
expressed <strong>in</strong> terms of the Hermitian product def<strong>in</strong>ed at C n+1 .<br />
Proposition 1.2.3 ([Gol99] page 76). Let x, y ∈ CK n (ɛ), ɛ ≠ 0, and let d be the distance<br />
between the two given po<strong>in</strong>ts. If x ′ and y ′ are representatives of x and y, respectively, <strong>in</strong> the<br />
projective model, then the distance between the two po<strong>in</strong>ts is given by<br />
(cos ɛ d(x, y)) 2 = (x′ , y ′ )(y ′ , x ′ )<br />
(x ′ , x ′ )(y ′ , y ′ )<br />
where ( , ) denotes the Hermitian product <strong>in</strong> C n+1 def<strong>in</strong>ed at (1.2).<br />
1.2.5 Isometries<br />
Let us recall the def<strong>in</strong>ition of the matrix Lie group U(p, q).<br />
Def<strong>in</strong>ition 1.2.4. Let (x, y) = − ∑ p−1<br />
j=0 x jy j + ∑ n<br />
j=p x jy j be a Hermitian product <strong>in</strong> C n and<br />
p, q ∈ N ∪ {0} such that p + q = n + 1. Then it is def<strong>in</strong>ed<br />
U(p, q) = {A ∈ M n×n (C) | (Av, Aw) = (v, w) with v, w ∈ C n }.<br />
The matrix group U(n) co<strong>in</strong>cides with U(0, n), that is, we consider the standard Hermitian<br />
product on C n .<br />
The matrices of P U(n + 1) = U(n + 1)/(multiplication by scalars), if ɛ > 0 (resp. the<br />
matrices of P U(1, n) = U(1, n)/(multiplication by <strong>complex</strong> scalars), if ɛ < 0) act naturally on<br />
CK n (ɛ). Moreover, they preserve the metric def<strong>in</strong>ed <strong>in</strong> the model s<strong>in</strong>ce preserve the Hermitian<br />
product def<strong>in</strong>ed at C n+1 . Then, the matrices <strong>in</strong> P U(n + 1) (resp. P U(1, n)) are isometries of<br />
CP n (resp. CH n ).<br />
Proposition 1.2.5 ([Gol99] page 68).<br />
map <strong>in</strong> C n+1 .<br />
• Every isometry of CK n (ɛ) comes from a l<strong>in</strong>ear<br />
• The isometry group of CK n (ɛ) is P U ɛ (n) with<br />
⎧<br />
⎨ C n ⋊ U(n), if ɛ = 0,<br />
U ɛ (n) = U(n + 1) = U(0, n + 1), if ɛ > 0,<br />
⎩<br />
U(1, n), if ɛ < 0.<br />
(1.5)
12 Spaces of constant holomorphic curvature<br />
In order to unify the study of the <strong>complex</strong> <strong>space</strong> <strong>forms</strong>, <strong>in</strong>dependently of ɛ, we represent<br />
the elements <strong>in</strong> the group C n ⋊ U(n) as matrices<br />
( )<br />
1 0<br />
C n . (1.6)<br />
U(n)<br />
The transitivity of the isometry group at different levels is given <strong>in</strong> the follow<strong>in</strong>g proposition.<br />
Proposition 1.2.6 ([Gol99] page 70). The isometry group of CK n (ɛ) acts transitively<br />
• on the po<strong>in</strong>ts <strong>in</strong> CK n (ɛ),<br />
• on the unit tangent bundle. That is, given (p, v), (q, w) <strong>in</strong> the unit tangent bundle there<br />
exists an isometry σ such that σ(p) = q and dσ(v) = w.<br />
• on the holomorphic sections (see Def<strong>in</strong>ition 1.1.10).<br />
In the follow<strong>in</strong>g lemma, we give a basis of left-<strong>in</strong>variant <strong>forms</strong> of U(p, q). We will prove<br />
that these <strong>forms</strong> are also right-<strong>in</strong>variants.<br />
Lemma 1.2.7. Let A = (a 0 , . . . , a m ) ∈ U(p, q), with p, q, m ∈ N∪{0} such that p+q = m+1.<br />
A basis of left-<strong>in</strong>variant <strong>forms</strong> <strong>in</strong> U(p, q) is given by {Re(ϕ jk ), Im(ϕ jk ), Re(ϕ jj )}, 0 ≤ j ≤ k ≤<br />
m, j ≠ k where ϕ ij = (da i , a j ) and (x, y) = − ∑ p−1<br />
j=0 x jy j + ∑ m<br />
j=p x jy j <strong>in</strong> C m+1 .<br />
Proof. From Def<strong>in</strong>ition 1.2.4 of U(p, q) it follows that A ∈ U(p, q) if and only if A −1 = εA t ε<br />
where<br />
( )<br />
−Idp 0<br />
ε =<br />
. (1.7)<br />
0 Id q<br />
In order to f<strong>in</strong>d a basis of left-<strong>in</strong>variant <strong>forms</strong> we compute A −1 dA with A ∈ U(p, q). If we<br />
denote A = (a 0 , . . . , a m ), then<br />
⎛<br />
⎞<br />
(da 0 , −a 0 ) . . . (da m , −a 0 )<br />
A −1 dA = εA T .<br />
.<br />
εdA =<br />
⎜ (da 0 , −a p−1 ) . . . (da m , −a p−1 )<br />
= (ϕ ij ) ij . (1.8)<br />
⎟<br />
⎝ (da 0 , a p ) . . . (da m , a p ) ⎠<br />
(da 0 , a m ) . . . (da m , a m )<br />
Each entry of this matrix is a 1-form given by ϕ ij = ±(da i , a j ).<br />
Note that each a j is a m-tuple of <strong>complex</strong> numbers, so that the 1-<strong>forms</strong> ϕ ij are <strong>complex</strong>valued.<br />
In order to f<strong>in</strong>d a basis of left-<strong>in</strong>variant real-valued 1-<strong>forms</strong> from the entries of the former<br />
matrix we use the follow<strong>in</strong>g<br />
• (a j , a j ) = ±1 and differentiat<strong>in</strong>g<br />
0 = (a j , da j ) + (da j , a j ) = (da j , a j ) + (da j , a j ) = 2Re(da j , a j ).<br />
Thus, ϕ jj = −ϕ jj and each ϕ jj takes only imag<strong>in</strong>ary values.<br />
• (a j , a k ) = 0 if k ≠ j and differentiat<strong>in</strong>g<br />
0 = (da j , a k ) + (a j , da k ).<br />
Thus, {<br />
ϕjk = ϕ kj if j ∈ {0, . . . , p − 1} or k ∈ {0, . . . , p − 1}<br />
ϕ jk = −ϕ kj<br />
otherwise.
1.2 Projective model 13<br />
On the other hand, it follows directly from the def<strong>in</strong>ition of U(p, q) that its dimension is<br />
(p + q) 2 . Then, {Re(ϕ jk ), Im(ϕ jk ), Re(ϕ jj )}, 0 ≤ j ≤ k ≤ m, j ≠ k constitutes a basis of<br />
left-<strong>in</strong>variant 1-<strong>forms</strong> s<strong>in</strong>ce they generate all the <strong>space</strong> and there are (p + q) 2 <strong>forms</strong>.<br />
For ɛ = 0, if we denote the elements A ∈ C n ⋊ U(n) as the matrices of (1.8), we def<strong>in</strong>e the<br />
<strong>forms</strong><br />
ϕ ij = (da i , a j )<br />
with ( , ) the standard product <strong>in</strong> C n+1 .<br />
Def<strong>in</strong>ition 1.2.8. A group G is said to be unimodular if there exists a volume element of G<br />
left and right-<strong>in</strong>variant.<br />
Lemma 1.2.9. U(p + q) is a unimodular group.<br />
Proof. From Lemma 1.2.7 we have a basis of left-<strong>in</strong>variant <strong>forms</strong> of U(p, q). We prove that<br />
each of these <strong>forms</strong> is also right-<strong>in</strong>variant, i.e. it satisfies<br />
R ∗ Bϕ ij (A; v) = ϕ ij (A; v)<br />
∀A, B ∈ U(p, q), v ∈ T A U(p, q).<br />
We use the expression ϕ ij = ±(da i , a j ) and we denote by a k (A) the map tak<strong>in</strong>g the k-th<br />
column of a matrix A. Then, if A, B ∈ U(p, q) and v ∈ T A U(p, q)<br />
RBϕ ∗ ij (A; v) = ϕ ij (R B (A); d(R B )(v)) = ±(da i (d(R B )(v)), a j (R B (A))) = ±(da i (vB), a j (AB))<br />
( m∑ ∑p−1<br />
m<br />
= ± − vi k b r k al j br l +<br />
∑ m∑<br />
)<br />
vi k b r k al j br l<br />
(<br />
= ± −<br />
(<br />
= ±<br />
k,l=0 r=0<br />
p−1<br />
∑<br />
k,l=0<br />
δ kl v k i a l j +<br />
m ∑<br />
k,l=p<br />
k,l=0 r=p<br />
δklv k i a l j<br />
∑p−1<br />
m<br />
− vi k a k j + ∑<br />
)<br />
vi k a k j<br />
= ±(da i (v), a j (A)) = ϕ ij (A; v).<br />
k=0<br />
k=p<br />
Then, U(p, q) is a unimodular group s<strong>in</strong>ce the volume element obta<strong>in</strong>ed from the product of the<br />
<strong>forms</strong> ϕ ij is left-<strong>in</strong>variant and right-<strong>in</strong>variant (it is a product of <strong>forms</strong> with this property).<br />
)<br />
1.2.6 Structure of homogeneous <strong>space</strong><br />
Def<strong>in</strong>ition 1.2.10. Let (M, g) be a Riemannian manifold. If given any two po<strong>in</strong>ts x, y ∈ M<br />
there exists an isometry σ of M such that σ(x) = y, then M is a homogeneous <strong>space</strong>. That is,<br />
a Riemannian manifold is homogeneous if it is a homogeneous <strong>space</strong> of its isometry group.<br />
By Proposition 1.2.6 we have that <strong>complex</strong> <strong>space</strong> <strong>forms</strong> are homogeneous <strong>space</strong>s. It will<br />
be <strong>in</strong>terest<strong>in</strong>g to represent them as a quotient of Lie groups.<br />
Proposition 1.2.11 ([War71] Theorem 3.62 page 123). Let η : G × M → M be a transitive<br />
action of the Lie group G over the manifold M. Let m 0 ∈ M and H the isotropy group of m 0 .<br />
Then, the map<br />
β : G/H −→ M<br />
gH ↦→ η(g, m 0 )<br />
is a diffeomorphism.
14 Spaces of constant holomorphic curvature<br />
The Lie group P U ɛ (n) acts transitively over CK n (ɛ) and the isotropy group of a po<strong>in</strong>t <strong>in</strong><br />
CK n (ɛ), for each ɛ, is isomorphic to P (U(1) × U(n)), a closed Lie subgroup of P U ɛ (n). Thus,<br />
we can represent CK n (ɛ) as a quotient of Lie groups<br />
CK n (ɛ) ∼ = P U ɛ (n)/P (U(1) × U(n)) ∼ = U ɛ (n)/(U(1) × U(n)),<br />
where the first diffeomorphism is given by x ↦→ g · P (U(1) × U(n)) with g ∈ P U ɛ (n) such that,<br />
if x 0 ∈ CK n (ɛ) is fixed then g(x 0 ) = x.<br />
Def<strong>in</strong>ition 1.2.12. A Riemannian manifold is a 2-po<strong>in</strong>t homogeneous <strong>space</strong> if the isometry<br />
group of the manifold acts transitively <strong>in</strong> the unit tangent bundle.<br />
Thus, <strong>complex</strong> <strong>space</strong> <strong>forms</strong> are 2-po<strong>in</strong>t homogeneous <strong>space</strong>s. Real <strong>space</strong> <strong>forms</strong> are also 2-<br />
po<strong>in</strong>t homogeneous <strong>space</strong>s but, they are also 3-po<strong>in</strong>t homogeneous <strong>space</strong>s, that is, the isometry<br />
group acts transitively for triplets of a po<strong>in</strong>t and two orthonormal vectors <strong>in</strong> the tangent <strong>space</strong><br />
of the po<strong>in</strong>t. Complex <strong>space</strong> <strong>forms</strong> (ɛ ≠ 0) cannot be 3-po<strong>in</strong>t homogeneous <strong>space</strong>s s<strong>in</strong>ce the<br />
sectional curvature is preserved by isometries and the sectional curvature is not constant.<br />
1.3 Mov<strong>in</strong>g frames<br />
Def<strong>in</strong>ition 1.3.1. Let U ⊂ M be an open set of a differentiable manifold. An orthonormal<br />
mov<strong>in</strong>g frame of CK n (ɛ) def<strong>in</strong>ed at U is a map g 0 : U → CK n (ɛ) together with a collection of<br />
g i : U → T CK n (ɛ) (i ∈ {1, . . . , 2n}) such that 〈g i , g j 〉 ɛ = δ ij where 〈 , 〉 ɛ denotes the Hermitian<br />
product of CK n (ɛ) (def<strong>in</strong>ed at (1.4)) and π : T CK n (ɛ) → CK n (ɛ) is the canonical projection.<br />
Def<strong>in</strong>ition 1.3.2. Let V be a 2n-dimensional real vector <strong>space</strong> endowed with a <strong>complex</strong><br />
structure J. An orthonormal basis {v 1 , v 2 , . . . , v 2n } is said to be a J-basis if v 2i = Jv 2i−1 for<br />
every i ∈ {1, . . . , n}.<br />
We denote by {e 1 , e 1 = Je 1 , . . . , e n , e n = Je n } the J-bases of V , and by {ω 1 , ω 1 , . . . , ω n , ω n }<br />
the dual basis of a J-basis.<br />
Remark 1.3.3. A J-basis is a special type of an orthonormal basis <strong>in</strong> a real vector <strong>space</strong> with<br />
an almost <strong>complex</strong> structure J.<br />
Def<strong>in</strong>ition 1.3.4. An orthonormal mov<strong>in</strong>g frame of CK n (ɛ) such that vectors {g 1 (p), . . . , g 2n (p)}<br />
constitute a J-basis for all x ∈ U, is called a J-mov<strong>in</strong>g frame.<br />
J-mov<strong>in</strong>g frames <strong>in</strong> CK n (ɛ) play an important role s<strong>in</strong>ce they are <strong>in</strong> correspondence with<br />
the elements of the isometry group of CK n (ɛ).<br />
Consider F(CK n (ɛ)) the bundle of J-mov<strong>in</strong>g frames of CK n (ɛ), constituted by J-mov<strong>in</strong>g<br />
frames (g 0 ; g 1 , Jg 1 , . . . , g n , Jg n ) with g 0 ∈ CK n (ɛ) and {g 1 , Jg 1 , . . . , g n , Jg n } a J-basis of<br />
T g0 CK n (ɛ).<br />
Proposition 1.3.5. The bundle of J-mov<strong>in</strong>g frames F(CK n (ɛ)) is identified with the isometry<br />
group of CK n (ɛ).<br />
Proof. We study the case ɛ ≠ 0. Let A ∈ U ɛ (n). By def<strong>in</strong>ition (1.5) of U ɛ (n) we have that A is<br />
an (n + 1) × (n + 1) matrix with <strong>complex</strong> entries and such that its columns {a 0 , . . . , a n } satisfy<br />
⎧<br />
⎨<br />
⎩<br />
(a 0 , a 0 ) = sign(ɛ)1,<br />
(a 0 , a i ) = 0, i ∈ {1, . . . , n},<br />
(a i , a j ) = δ ij i, j ∈ {1, . . . , n},<br />
where ( , ) denotes the Hermitian product <strong>in</strong> C n+1 def<strong>in</strong>ed at (1.2).<br />
(1.9)
1.3 Mov<strong>in</strong>g frames 15<br />
From the first property <strong>in</strong> the former list, we can take a 0 as a representative of g 0 = π(a 0 ) ∈<br />
CK n (ɛ).<br />
From the second property of (1.9) we have that a i satisfies the condition of be<strong>in</strong>g a vector<br />
<strong>in</strong> T a0 H (cf. Section 1.2.2). Moreover, Re(Ja 0 , a i ) = Im(a 0 , a i ) = 0 and Re(Ja 0 , Ja i ) =<br />
Re(a 0 , a i ) = 0. Let us consider g i := dπ(a i ) and Jg i := dπ(Ja i ) where π denotes the projection<br />
def<strong>in</strong>ed at (1.3).<br />
From the third condition <strong>in</strong> (1.9), vectors {g 1 , Jg 1 , . . . , g n , Jg n } constitute a J-basis of the<br />
tangent <strong>space</strong> at g 0 .<br />
Reciprocally, given a J-mov<strong>in</strong>g frame {g 0 ; g 1 , Jg 1 , . . . , g n , Jg n } def<strong>in</strong>ed on an open set, we<br />
can def<strong>in</strong>e a matrix of U ɛ (n) (with the entries depend<strong>in</strong>g cont<strong>in</strong>uously on a parameter) just<br />
tak<strong>in</strong>g as the first column the representative a 0 of g 0 with norm sign(ɛ)1. For the other columns<br />
we consider the horizontal lift of g j at a 0 . As {g 1 , Jg 1 , . . . , g n , Jg n } is, <strong>in</strong> each po<strong>in</strong>t g, a J-basis<br />
and we choose the horizontal lift, the columns of the constructed matrix verify the conditions<br />
<strong>in</strong> (1.9) and are <strong>in</strong> U ɛ (n).<br />
Def<strong>in</strong>ition 1.3.6. The unit tangent bundle of CK n (ɛ), denoted by S(CK n (ɛ)), is def<strong>in</strong>ed as<br />
S(CK n (ɛ)) =<br />
⋃<br />
p∈CK n (ɛ)<br />
T ′ pCK n (ɛ)<br />
where T ′ pCK n (ɛ) denotes the sphere of unit vectors <strong>in</strong> the tangent vector <strong>space</strong> of CK n (ɛ) at<br />
p.<br />
In Lemma 1.2.7 we def<strong>in</strong>ed the <strong>in</strong>variant <strong>forms</strong> {ϕ ij } of U ɛ (n) as<br />
ϕ ij (A; ·) = (da i (·), a j )<br />
where A = (a 0 , . . . , a n ) ∈ U ɛ (n). As <strong>forms</strong> {ϕ ij } takes <strong>complex</strong> values we consider<br />
ϕ jk = α jk + iβ jk (1.10)<br />
Us<strong>in</strong>g the identification between U ɛ (n) and F(CK n (ɛ)), we can consider <strong>forms</strong> {ϕ ij } as<br />
<strong>forms</strong> of F(CK n (ɛ)).<br />
On the other hand, consider the canonical projections<br />
and local sections<br />
π 1<br />
F(CK n (ɛ)) −→ S(CK n (ɛ)) −→ CK n (ɛ)<br />
(g; g 1 , . . . , Jg n ) ↦→ (g, g 1 ) ↦→ g<br />
CK n (ɛ) ⊃ U s 2<br />
−→ S(CK n (ɛ)) ⊃ V s 1<br />
−→ F(CK n (ɛ)).<br />
Us<strong>in</strong>g the <strong>forms</strong> ϕ ij def<strong>in</strong>ed <strong>in</strong> F(CK n (ɛ)) and the previous local sections, we def<strong>in</strong>e the<br />
follow<strong>in</strong>g local <strong>in</strong>variant <strong>forms</strong> <strong>in</strong> S(CK n (ɛ))<br />
and the local <strong>in</strong>variant <strong>forms</strong> <strong>in</strong> CK n (ɛ)<br />
s ∗ 1(ϕ ij ), s ∗ 1(α ij ) and s ∗ 1(β ij )<br />
s ∗ 2s ∗ 1(ϕ ij ), s ∗ 2s ∗ 1(α ij ) and s ∗ 2s ∗ 1(β ij ).<br />
Lemma 1.3.7. Forms s ∗ 1 α 01, s ∗ 1 β 01 and s ∗ 1 β 11 are global <strong>forms</strong> <strong>in</strong> S(CK n (ɛ)).<br />
π 2
16 Spaces of constant holomorphic curvature<br />
Proof. If V ∈ T (p,v) (S(CK n (ɛ))) then<br />
s ∗ 1α 01 (V ) (p,v) = Re((dπ 2 (V ), v)) = 〈dπ 2 (V ), v〉 ɛ<br />
s ∗ 1β 01 (V ) (p,v) = Im((dπ 2 (V ), v)),<br />
s ∗ 1β 11 (V ) (p,v) = Im((∇V, v)),<br />
where ∇ denotes the Levi-Civita connection def<strong>in</strong>ed by ∇ : T (SCK n (ɛ)) → T CK n (ɛ), from<br />
the Levi-Civita connection of CK n (ɛ). Inded, a vector V ∈ T (SCK n (ɛ)) is a tangent vector<br />
to a curve of unit vectors, and <strong>in</strong> each of them we can apply the Levi-Civita connection of<br />
CK n (ɛ).<br />
We denote by α, β, γ the <strong>forms</strong> s ∗ 1 α 01, s ∗ 1 β 01, s ∗ 1 β 11, respectively.<br />
Remark 1.3.8. The 1-form α co<strong>in</strong>cides with the standard contact form of the unit tangent<br />
bundle S(CK n (ɛ)).<br />
On the other hand, <strong>forms</strong> s ∗ 2 s∗ 1 (ϕ ij) co<strong>in</strong>cide with <strong>forms</strong> φ ij of CK n (ɛ) we def<strong>in</strong>e <strong>in</strong> the<br />
follow<strong>in</strong>g.<br />
Let {g; g 1 , Jg 1 , . . . , g n , Jg n } be a J-mov<strong>in</strong>g frame on CK n (ɛ). As <strong>in</strong> the tangent <strong>space</strong> of<br />
each po<strong>in</strong>t g, {g 1 , Jg 1 , . . . , g n , Jg n } def<strong>in</strong>es a J-basis, we can consider the vectors {g 1 , . . . , g n }<br />
as <strong>complex</strong> vectors. Then, the follow<strong>in</strong>g differential <strong>forms</strong> are well-def<strong>in</strong>ed<br />
φ j (·) = (dg(·), g j ) ɛ and φ jk (·) = (∇g j (·), g k ) ɛ (1.11)<br />
where j, k ∈ {1, ..., n}, and ∇ denotes the Levi-Civita of CK n (ɛ) (i.e. we consider g j as a real<br />
vector, we apply the Levi-Civita connection and we consider the result aga<strong>in</strong> as a <strong>complex</strong><br />
vector). Note that the differential <strong>forms</strong> φ j and φ jk are <strong>complex</strong> valued. We denote<br />
φ j = α j + iβ j (1.12)<br />
φ jk = α jk + iβ jk .<br />
At Chapter 3, we work with orthonormal mov<strong>in</strong>g frames not necessarily J-mov<strong>in</strong>g frames.<br />
Analogously, if {g; g 1 , g 2 , . . . , g 2n−1 , g 2n } is a mov<strong>in</strong>g frame on CK n (ɛ), we def<strong>in</strong>e the dual and<br />
connection <strong>forms</strong> for this mov<strong>in</strong>g frame. We denote<br />
ω j (·) = 〈dg(·), g j 〉 ɛ and ω jk (·) = 〈∇g j (·), g k 〉 ɛ (1.13)<br />
with j, k ∈ {1, ..., 2n}, 〈 , 〉 ɛ the Hermitian product def<strong>in</strong>ed on CK n (ɛ) (see (1.4)), and ∇ the<br />
Levi-Civita connection of CK n (ɛ).<br />
Note that the differential <strong>forms</strong> {α j , β j } are a particular case of <strong>forms</strong> {ω j }: they are<br />
obta<strong>in</strong>ed if we consider a J-mov<strong>in</strong>g frame.<br />
Notation 1.3.9. Along this work we use <strong>in</strong>variant <strong>forms</strong> def<strong>in</strong>ed at CK n (ɛ), S(CK n (ɛ)) or<br />
F(CK n (ɛ)), but we denote all of them by ϕ ij , α ij , β ij , without the pull-back of the sections,<br />
if it is clear by the context.<br />
Def<strong>in</strong>ition 1.3.10. Given a doma<strong>in</strong> Ω ⊂ CK n (ɛ) we def<strong>in</strong>e the unit normal bundle of ∂Ω by<br />
N(Ω)={(p, v) : p ∈ ∂Ω, v such that 〈v, w〉 ɛ ≥ 0 ∀ w tangent to a curve at Ω by p and ||v|| ɛ =1}.<br />
Remark 1.3.11. The ma<strong>in</strong> results of this work, given at Chapters 3 and 4, have as a hypothesis<br />
that the doma<strong>in</strong> Ω ⊂ CK n (ɛ) which we take is compact with C 2 boundary. We denote by<br />
regular doma<strong>in</strong> a doma<strong>in</strong> satisfy<strong>in</strong>g these hypothesis. We suppose that doma<strong>in</strong>s are regular <strong>in</strong><br />
order to simplify the arguments and to use techniques of differential <strong>geometry</strong> (for <strong>in</strong>stance,<br />
to have a well-def<strong>in</strong>ed second fundamental form <strong>in</strong> the whole boundary of the doma<strong>in</strong>). These<br />
hypothesis can be relaxed s<strong>in</strong>ce most of the used results, ma<strong>in</strong>ly <strong>in</strong> valuations (see Chapter 2),<br />
are known for a more general class of doma<strong>in</strong>s (cf. [Ale07a]).
1.4 Submanifolds 17<br />
Lemma 1.3.12. Let Ω ⊂ CK n (ɛ) be a regular doma<strong>in</strong>. Forms α and dα vanishes at N(Ω) ⊂<br />
S(CK n (ɛ)).<br />
Proof. Let V ∈ T (p,v) Ω. Then, α(V ) (p,N) = 〈dπ 2 (V ), N〉 ɛ = 0 s<strong>in</strong>ce dπ 2 (V ) is a tangent vector<br />
tangent at ∂Ω at po<strong>in</strong>t p.<br />
In order to prove that the 2-form dα vanishes at the unit normal bundle, we consider the<br />
<strong>in</strong>clusion of the unit normal bundle to the unit tangent bundle i : N(Ω) → S(Ω). Us<strong>in</strong>g that<br />
the differential map commutes with the <strong>in</strong>clusion we have the result<br />
dα| N(Ω) = (i ∗ ◦ d)(α) = (d ◦ i ∗ )(α) = d(i ∗ α) = d(0) = 0.<br />
1.4 Submanifolds<br />
1.4.1 Totally geodesic submanifolds<br />
Def<strong>in</strong>ition 1.4.1. Let M be a Riemannian manifold. A submanifold N ⊂ M is totally geodesic<br />
if every geodesic <strong>in</strong> the submanifold N is also a geodesic <strong>in</strong> M.<br />
As C n is metrically equivalent to R 2n , totally geodesic submanifolds <strong>in</strong> C n co<strong>in</strong>cide with<br />
the ones <strong>in</strong> R 2n . For the other <strong>complex</strong> <strong>space</strong> <strong>forms</strong>, the totally geodesic submanifolds are<br />
classified.<br />
Def<strong>in</strong>ition 1.4.2. Let V be a real vector <strong>space</strong> of dimension 2n endowed with an almost <strong>complex</strong><br />
structure J compatible with a scalar product 〈 , 〉. It is said that vectors {e 1 , . . . , e m } expand<br />
a <strong>complex</strong> sub<strong>space</strong> if the <strong>space</strong> generated by these vectors is J-<strong>in</strong>variant, i.e. J(span{e 1 , . . . , e m }) =<br />
span{e 1 , . . . , e m }.<br />
It is said that a submanifold of a <strong>complex</strong> manifold is <strong>complex</strong> if at each po<strong>in</strong>t, the tangent<br />
<strong>space</strong> of the submanifold is <strong>complex</strong> sub<strong>space</strong> of the tangent <strong>space</strong> of the manifold.<br />
Def<strong>in</strong>ition 1.4.3. Let V be a real vector <strong>space</strong> of dimension 2n endowed with an almost<br />
<strong>complex</strong> structure J compatible with a scalar product 〈 , 〉. It is said that vectors {e 1 , . . . , e m }<br />
expand a totally real sub<strong>space</strong> if<br />
〈e i , Je j 〉 = 0, ∀i, j ∈ {1, . . . , m}.<br />
It is said that a submanifold of a <strong>complex</strong> manifold is totally real if at each po<strong>in</strong>t, the tangent<br />
<strong>space</strong> of the submanifold is a totally real sub<strong>space</strong> of the tangent <strong>space</strong> of the manifold.<br />
Theorem 1.4.4 ([Gol99] pages 75 and 80). Let z ∈ CK n (ɛ).<br />
1. If L ⊂ T z CK n (ɛ) is a <strong>complex</strong> vector sub<strong>space</strong> with <strong>complex</strong> dimension r, then there<br />
exists a unique complete <strong>complex</strong> totally geodesic submanifold through z and tangent to<br />
L at z.<br />
2. If L ⊂ T z CK n (ɛ) is a totally real vector <strong>space</strong> of real dimension k, then there exists a<br />
unique complete totally geodesic totally real submanifold through z and tangent to L at<br />
z.<br />
Def<strong>in</strong>ition 1.4.5. The <strong>complex</strong> submanifold def<strong>in</strong>ed at 1. <strong>in</strong> the previous theorem is called<br />
<strong>complex</strong> r-plane, and denoted by L r .<br />
The totally real submanifold def<strong>in</strong>ed at 2. <strong>in</strong> the previous theorem is called totally real<br />
k-plane, and denoted by L R k .
18 Spaces of constant holomorphic curvature<br />
In the projective model, <strong>complex</strong> r-planes are obta<strong>in</strong>ed from the projection of a sub<strong>space</strong><br />
F ⊂ C n+1 <strong>in</strong>tersection CK n (ɛ). The sub<strong>space</strong> F is the (r + 1)-dimensional <strong>complex</strong> vector<br />
sub<strong>space</strong> spanned by a representative z ′ of z = π(z ′ ) and by the horizontal lift of vectors <strong>in</strong><br />
L ⊂ T z CK n (ɛ) (cf. [Gol99, Section 3.1.4]).<br />
Analogously, totally real k-planes are obta<strong>in</strong>ed from the projection of F R ⊂ C n+1 <strong>in</strong>tersection<br />
CK n (ɛ), where F R is the (k + 1)-dimensional real vector sub<strong>space</strong> spanned by a<br />
representative z ′ of z = π(z ′ ) and by the horizontal lift of vectors <strong>in</strong> L ⊂ T z CK n (ɛ).<br />
Theorem 1.4.6 ([Gol99] p. 82). The unique complete totally geodesic submanifolds <strong>in</strong> CK n (ɛ)<br />
are the <strong>complex</strong> r-planes, r ∈ {1, . . . , n − 1} and the totally real k-planes, k ∈ {1, . . . , n}.<br />
Corollary 1.4.7. In CK n (ɛ), ɛ ≠ 0, there are not totally geodesic (real) hypersurfaces.<br />
That is, it does not exist the equivalent hypersurface to a hyperplane <strong>in</strong> a real <strong>space</strong> form.<br />
The more reasonable substitutes of hyperplanes are the so-called bisectors, which we study on<br />
page 82.<br />
Theorem 1.4.6 will be important along this work s<strong>in</strong>ce it will be <strong>in</strong>terest<strong>in</strong>g to know which<br />
totally geodesic submanifolds can be taken <strong>in</strong> a <strong>complex</strong> <strong>space</strong> form as a substitutes of (totally<br />
geodesic) planes <strong>in</strong> real <strong>space</strong> <strong>forms</strong>.<br />
1.4.2 Geodesic balls<br />
A geodesic ball <strong>in</strong> a Riemannian manifold is the set of po<strong>in</strong>ts equidistant from a fixed po<strong>in</strong>t<br />
called center.<br />
In real <strong>space</strong> <strong>forms</strong>, geodesic balls are totally umbilical real hypersurfaces, i.e. the second<br />
fundamental form is, at every po<strong>in</strong>t, a multiple of the identity and the same multiple for every<br />
po<strong>in</strong>t.<br />
This fact does not hold <strong>in</strong> <strong>complex</strong> projective and hyperbolic <strong>space</strong>s. Moreover, <strong>in</strong> these<br />
<strong>space</strong>s, there are no totally umbilical real hypersurface.<br />
Proposition 1.4.8 ([Mon85]). The pr<strong>in</strong>cipal curvatures of a sphere of radius r <strong>in</strong> CK n (ɛ),<br />
ɛ ≠ 0 are<br />
i) 2 cot ɛ (2r) with multiplicity 1 and pr<strong>in</strong>cipal direction −JN (where N denotes the <strong>in</strong>ward<br />
normal vector to the sphere),<br />
ii) cot ɛ (r) with multiplicity 2n − 2.<br />
Recall that cos ɛ , s<strong>in</strong> ɛ denote the generalized trigonometric functions def<strong>in</strong>ed at Notation<br />
1.2.2.<br />
Along this work, we use the value of the mean curvature <strong>in</strong>tegrals for a geodesic ball of<br />
radius R <strong>in</strong> CK n (ɛ) (cf. Def<strong>in</strong>ition 2.1.4).<br />
From the previous proposition we have that the symmetric elementary functions are<br />
⎧<br />
⎪⎨<br />
⎪⎩<br />
σ 0 = 1<br />
) −1<br />
( (2n−2<br />
σ i = ( 2n−1<br />
i<br />
σ 2n−1 = 2 cot 2n−2<br />
ɛ<br />
i<br />
)<br />
cot<br />
i<br />
ɛ (R) + ( 2n−2<br />
i−1<br />
(R) cot ɛ (2R).<br />
) )<br />
2 cot<br />
i−1(R) cot ɛ (2R)<br />
By a straightforward computation, we obta<strong>in</strong> that the expression of the mean curvature <strong>in</strong>tegrals<br />
is<br />
and<br />
⎧<br />
⎪⎨<br />
⎪⎩<br />
M 0 = vol(∂B R ) =<br />
2π<br />
M i =<br />
n<br />
(2n−1)(n−1)!<br />
M 2n−1 =<br />
2πn<br />
(n−1)! (cos2n ɛ<br />
2πn<br />
(n−1)! s<strong>in</strong>2n−1 ɛ<br />
((2n + i − 1) cosi+1 ɛ<br />
(R) + cos 2n−2<br />
ɛ<br />
(R) cos ɛ (R)<br />
(R) s<strong>in</strong> 2n−i−1<br />
ɛ<br />
(R) s<strong>in</strong> 2 ɛ(R))<br />
vol(B R ) = πn<br />
n! (s<strong>in</strong> ɛ(R)) 2n .<br />
ɛ<br />
(R) − i cos i−1 (R) s<strong>in</strong> 2n−i−1 (R))<br />
ɛ<br />
ɛ
1.5 Space of <strong>complex</strong> r-planes 19<br />
1.5 Space of <strong>complex</strong> r-planes<br />
We denote the <strong>space</strong> of all <strong>complex</strong> r-planes <strong>in</strong> CK n (ɛ) by L C r and the <strong>space</strong> of all totally real<br />
r-planes <strong>in</strong> CK n (ɛ) by L R r (cf. Def<strong>in</strong>ition 1.4.5).<br />
We shall use that L C r is a homogeneous <strong>space</strong> with respect to the isometry group of CK n (ɛ).<br />
In order to prove this fact, we need the follow<strong>in</strong>g result.<br />
Lemma 1.5.1. The group of isometries of CK n (ɛ) acts transitively on the J-bases.<br />
Proof. We study the case ɛ ≠ 0. Fix the canonical J-basis {e 1 , Je 1 , . . . , e n , Je n } at e 0 ∈<br />
CK n (ɛ). It is enough to prove that given another J-basis {g 1 , Jg 1 , . . . , g n , Jg n } of T g0 CK n (ɛ),<br />
there exists an isometry ρ which takes this J-basis to the fixed one.<br />
Take as isometry ρ ∈ U ɛ (n) the matrix with columns (˜g 0 , ˜g 1 , . . . , ˜g n ) where ˜g 0 is a representative<br />
of g 0 with norm ɛ and ˜g i is the horizontal lift of g i at ˜g 0 . In the same way as <strong>in</strong> the<br />
proof of Proposition 1.3.5 we have that ρ is a matrix <strong>in</strong> U ɛ (n). Moreover, it carries the fixed<br />
J-basis to the given one.<br />
From the previous lemma we get<br />
Lemma 1.5.2. The <strong>space</strong> of <strong>complex</strong> r-planes is a homogeneous <strong>space</strong> with respect to the<br />
isometry group of CK n (ɛ).<br />
Proof. We def<strong>in</strong>e a J-basis {e 1 , Je 1 , . . . , e r , Je r } of the tangent <strong>space</strong> <strong>in</strong> any po<strong>in</strong>t of a <strong>complex</strong><br />
r-plane. Complet<strong>in</strong>g this J-basis to a J-basis of the whole <strong>space</strong> CK n (ɛ), and apply<strong>in</strong>g the<br />
previous lemma we get the result.<br />
In order to study <strong>in</strong>tegral <strong>geometry</strong> <strong>in</strong> CK n (ɛ), it is necessary that the <strong>space</strong> L C r admits an<br />
<strong>in</strong>variant density under the isometries of CK n (ɛ). In general, the absolute value of a form of<br />
maximum degree is called a density. By the follow<strong>in</strong>g lemma, it is enough to prove that L C r is<br />
the quotient of unimodular groups.<br />
Lemma 1.5.3 ([San04]). If G, H are unimodular groups, then G/H admits an <strong>in</strong>variant<br />
density.<br />
The isotropy group of a <strong>complex</strong> r-plane is isomorphic to<br />
U ɛ (r) × U(n − r) =<br />
{<br />
( A 0<br />
X ∈ M (n+1)×(n+1) (C) : X =<br />
0 B<br />
)<br />
}<br />
, A ∈ U ɛ (r), B ∈ U(n − r)<br />
(1.14)<br />
s<strong>in</strong>ce these matrices (and only these) leave <strong>in</strong>variant a <strong>complex</strong> r-plane and its orthogonal.<br />
Then,<br />
L C r ∼ = U ɛ (n)/(U ɛ (r) × U(n − r)).<br />
The group U ɛ (n), by Lemma 1.2.9, is unimodular, and U ɛ (r)×U(n−r) is also a unimodular<br />
group. Thus, there exists an <strong>in</strong>variant density <strong>in</strong> the quotient <strong>space</strong>, that is, <strong>in</strong> the <strong>space</strong> of<br />
<strong>complex</strong> r-planes.<br />
The follow<strong>in</strong>g result give a method to obta<strong>in</strong> explicitly the density <strong>in</strong> the quotient <strong>space</strong>.<br />
Theorem 1.5.4 ([San04] page 147). Let G be a Lie group with dimension n and H a closed<br />
subgroup of G with dimension n − m. Then, G/H is a homogeneous <strong>space</strong> with dimension m.<br />
Let ˜ω be the m-form obta<strong>in</strong>ed from the product of all <strong>in</strong>variant 1-<strong>forms</strong> of G such that they<br />
vanish on H. Then, there exists an <strong>in</strong>variant density ω <strong>in</strong> G/H if and only if d˜ω = 0. In<br />
this case, ˜ω is the pull-back of ω for the canonical projection of G at G/H (up to constants<br />
factors).
20 Spaces of constant holomorphic curvature<br />
Proposition 1.5.5. Let π : U ɛ (n) −→ U ɛ (n)/(U ɛ (r) × U(n − r)) ∼ = L C r . If dL r is an <strong>in</strong>variant<br />
density of L C r then<br />
∧<br />
π ∗ dL r = ϕ ij ∧ ϕ ij<br />
i=0,...,r<br />
j=r+1,...,n<br />
or, equivalently,<br />
π ∗ dL r =<br />
∧<br />
i=0,...,r<br />
j=r+1,...,n<br />
α ij ∧ β ij ,<br />
where {ϕ ij , α ij , β ij } are the <strong>forms</strong> def<strong>in</strong>ed at Lemma 1.2.7 and at (1.11).<br />
Proof. The result is known for C n , see [San04].<br />
For the other <strong>complex</strong> <strong>space</strong> <strong>forms</strong>, we note that the (real) dimension of the <strong>space</strong> of<br />
<strong>complex</strong> r-plans co<strong>in</strong>cides with the degree of π ∗ dL r (and it is equal to 2(r + 1)(n − r)). Thus,<br />
it suffices to prove that each form ϕ ij with i ∈ {0, . . . , r}, j ∈ {r + 1, . . . , n} vanishes over<br />
U ɛ (n) × U(n − r). But, from the form of matrices <strong>in</strong> U ɛ (n) × U(n − r) given at (1.14), the<br />
result follows immediately.<br />
Let us give an example of mov<strong>in</strong>g frames <strong>in</strong> the <strong>space</strong> of <strong>complex</strong> r-planes us<strong>in</strong>g Def<strong>in</strong>ition<br />
1.3.1, which will be used <strong>in</strong> the next section.<br />
Let us take as the open set U ⊂ M (see Def<strong>in</strong>ition 1.3.1) an open set <strong>in</strong> L C r . An orthonormal<br />
frame is given by<br />
g : U ⊂ L C r −→ CK n (ɛ)<br />
L r ↦→ p ∈ L r<br />
and<br />
g i : U ⊂ L C r −→ T CK n (ɛ)<br />
L r ↦→ v i ∈ T g(Lr)L r<br />
, (1.15)<br />
i ∈ {1, . . . , 2r}, such that 〈v i , v j 〉 ɛ = δ ij . It will be <strong>in</strong>terest<strong>in</strong>g to consider that Jg 2k−1 (L r ) =<br />
g 2k (L r ), k ∈ {1, . . . , r}, that is, vectors {g 1 (L r ), . . . , g 2r (L r )} constitute a J-basis at g(L r ). By<br />
abuse of notation, we denote g(L r ) by g and g i (L r ) by g i .<br />
Remark 1.5.6. From the correspondence between U ɛ (n) and the bundle of J-mov<strong>in</strong>g frames<br />
F(CK n (ɛ)), and between U ɛ (n)/(U ɛ (r)×U(n−r)) and L C r we have that {p; g 1 (L r ), . . . , g 2r (L r )}<br />
are sections of U ɛ (n) → U ɛ (n)/(U ɛ (r) × U(n − r)).<br />
1.5.1 Expression for the <strong>in</strong>variant density <strong>in</strong> terms of a parametrization<br />
Sometimes it is <strong>in</strong>terest<strong>in</strong>g to have a more geometric expression for the <strong>in</strong>variant density of<br />
<strong>complex</strong> r-planes. For example, Santaló proved<br />
Proposition 1.5.7 ([San04]). The <strong>in</strong>variant density of the <strong>space</strong> of totally geodesic planes <strong>in</strong><br />
a <strong>space</strong> form of constant sectional curvature k is<br />
dL r = cos r k (ρ)dx n−r ∧ dL (n−r)[O] (1.16)<br />
where dx n−r is the volume element of the (n − r)-plane orthogonal to L r conta<strong>in</strong><strong>in</strong>g the orig<strong>in</strong><br />
O and dL (n−r)[O] is the volume element of the Grassmannian of (n − r)-planes conta<strong>in</strong><strong>in</strong>g the<br />
orig<strong>in</strong>.<br />
Now, we give a similar expression <strong>in</strong> <strong>complex</strong> <strong>space</strong> <strong>forms</strong>, for the density of <strong>complex</strong><br />
r-planes <strong>in</strong> CK n (ɛ).
1.5 Space of <strong>complex</strong> r-planes 21<br />
Proposition 1.5.8. The <strong>in</strong>variant density of the <strong>space</strong> of <strong>complex</strong> r-planes <strong>in</strong> a <strong>space</strong> form<br />
of constant holomorphic curvature 4ɛ is<br />
dL r = cos 2r<br />
ɛ (ρ)dx n−r ∧ dL (n−r)[O]<br />
where dx n−r is the volume element of the <strong>complex</strong> (n−r)-plane orthogonal to L r conta<strong>in</strong><strong>in</strong>g the<br />
orig<strong>in</strong> O and dL (n−r)[O] is the volume element of the Grassmannian of <strong>complex</strong> (n − r)-planes<br />
conta<strong>in</strong><strong>in</strong>g the orig<strong>in</strong>.<br />
Proof. In order to obta<strong>in</strong> the expression of the <strong>in</strong>variant density of L C r <strong>in</strong> terms of dL (n−r)[O]<br />
and dx n−r , we follow the same idea as <strong>in</strong> [San04] (where it is proved for the Euclidean and the<br />
real hyperbolic <strong>space</strong>). That is, we fix an adapted J-mov<strong>in</strong>g frame to the <strong>complex</strong> r-plane,<br />
def<strong>in</strong>ed <strong>in</strong> a neighborhood of the po<strong>in</strong>t at m<strong>in</strong>imum distance from the orig<strong>in</strong> O, then, by<br />
parallel translation, we translate this mov<strong>in</strong>g frame to the orig<strong>in</strong> O. F<strong>in</strong>ally, we relate both<br />
mov<strong>in</strong>g frames from the pull-back of a section.<br />
Denote by G = U ɛ (n) and by H = U ɛ (r)×U(n−r) the isotropy group of a <strong>complex</strong> r-plane.<br />
The projection π : G −→ G/H gives, with respect to a J-mov<strong>in</strong>g frame adapted to the<br />
<strong>complex</strong> r-plane and the <strong>forms</strong> def<strong>in</strong>ed at Lemma 1.2.7,<br />
π ∗ dL r =<br />
n∧<br />
j=r+1<br />
α j0 ∧ β j0<br />
∧<br />
i=1,...,r<br />
j=r+1,...,n<br />
α ji ∧ β ji .<br />
Let O ∈ CK n (ɛ) be the fixed orig<strong>in</strong> and let L r ∈ L C r . We denote by p(L r ) the po<strong>in</strong>t <strong>in</strong> L r<br />
at a m<strong>in</strong>imum distance from O.<br />
Let i be a local section of π. Then, π ◦ i = id and i ∗ π ∗ dL r = dL r , so that, we can obta<strong>in</strong><br />
the expression of dL r . From the identifications expla<strong>in</strong>ed <strong>in</strong> Remark 1.5.6, we take as a section<br />
i the def<strong>in</strong>ed by {p(L r ); g 1 , Jg 1 , . . . , g n , Jg n }, a J-mov<strong>in</strong>g frame def<strong>in</strong>ed <strong>in</strong> a neighborhood V<br />
of L r , adapted to L r and such that g r+1 is the tangent vector to the geodesic jo<strong>in</strong><strong>in</strong>g p(L r )<br />
and O. Denote by {g 1 , Jg 1 , . . . , g n , Jg n } the dual basis of {g 1 , Jg 1 , . . . , g n , Jg n } at g 0 . From<br />
Proposition 1.3.5, we consider the matrix <strong>in</strong> G correspond<strong>in</strong>g to this J-mov<strong>in</strong>g frame. Denote<br />
the columns of G also by (g 0 , g 1 , . . . , g n ), so that (g i ◦ i) denotes the i-th column of the matrix.<br />
Then, us<strong>in</strong>g the same notation as <strong>in</strong> (1.12), we have<br />
n∧<br />
i ∗ ( α j0 ∧ β j0 ) =<br />
j=r+1<br />
=<br />
n∏<br />
n∏<br />
α j0 (di) ∧ β j0 (di) = 〈dg 0 (di), g j 〉〈dg 0 (di), Jg j 〉<br />
j=r+1<br />
n∏<br />
j=r+1<br />
〈d(g 0 ◦ i), g j 〉〈d(g 0 ◦ i), Jg j 〉 = (g 0 ◦ i) ∗ (<br />
j=r+1<br />
j=r+1<br />
n∧<br />
g j ∧ Jg j )<br />
but (g 0 ◦ i) = p(L r ) and the previous form co<strong>in</strong>cides with the volume element of p <strong>in</strong> the<br />
sub<strong>space</strong> generated by {g r+1 , Jg r+1 , . . . , g n , Jg n }, which is a <strong>complex</strong> (n − r)-plane. Thus,<br />
i ∗ (<br />
n∧<br />
j=r+1<br />
α j0 ∧ β j0 ) = dx n−r .<br />
Let G ′ ⊂ G be the subgroup of all J-mov<strong>in</strong>g frames of G such that g r+1 is the tangent<br />
vector to the geodesic conta<strong>in</strong><strong>in</strong>g O and g 0 , and let G 0 ⊂ G be the subgroup of all J-mov<strong>in</strong>g<br />
frames with base po<strong>in</strong>t O. Let ρ be the distance from L r to O. Denote by s ρ the parallel<br />
translation from O to p(L r ) along the geodesic. Consider the follow<strong>in</strong>g maps<br />
π 1 : G ′ −→ G 0<br />
(g 0 ; g 1 , . . . , g n ) ↦→ (O; s −1<br />
ρ (g 1 ), . . . , s −1<br />
ρ (g n )) ,
22 Spaces of constant holomorphic curvature<br />
π 2 : G 0 −→ L C r[O]<br />
G 0 ↦→ <strong>complex</strong> r-plane conta<strong>in</strong><strong>in</strong>g O generated by g 1 , . . . , g r<br />
,<br />
π 3 : L C r −→ L C r[0]<br />
L r ↦→ ((L r ) ⊥ O )⊥ O<br />
where (L r ) ⊥ O denotes the orthogonal <strong>space</strong> to L r by O. Then, the follow<strong>in</strong>g diagram is commutative<br />
G ′ π −−−−→<br />
1<br />
G0<br />
⏐ ⏐<br />
π↓<br />
↓ π 2<br />
(1.17)<br />
L C r<br />
We def<strong>in</strong>e curves x ij <strong>in</strong> G 0 ⊂ G as follows<br />
π 3<br />
−−−−→ L<br />
C<br />
r[0]<br />
.<br />
x ij (t) = (O; g 1 , . . . , g i (t), Jg i (t), . . . , g j (t), Jg j (t) . . . , g n , Jg n )<br />
where g i (t) = cos(t)g i + s<strong>in</strong>(t)g j and g j (t) = − s<strong>in</strong>(t)g i + cos(t)g j and curves<br />
x j i (t) = (O; g 1, . . . , g i (t), Jg i (t), . . . , g j (t), Jg j (t), . . . , g n , Jg n )<br />
where g i (t) = cos(t)g i + s<strong>in</strong>(t)Jg j and g j (t) = − s<strong>in</strong>(t)Jg i − cos(t)g j .<br />
From the local section i, we have<br />
i ∗ α ji = i ∗ ((π ∗ 1 ◦ s ∗ )(α ji )) = (π 1 ◦ i) ∗ (s ∗ ρα ji ),<br />
i ∗ β ji = i ∗ ((π ∗ 1 ◦ s ∗ )(β ji )) = (π 1 ◦ i) ∗ (s ∗ ρβ ji ).<br />
Thus, we have to study (s ∗ ρα ij )(ẋ kl ) and (s ∗ ρα ij )(ẋ l k ) (and the same for β ij). Then, we need<br />
(g l ◦ s ρ )(x ij ) s<strong>in</strong>ce<br />
Note that<br />
(s ∗ ρα ij )(ẋ kl ) = α ij (ds ρ (ẋ kl )) = 〈g j<br />
∣<br />
∣sρ(x<br />
kl (t)) , d(g i ◦ s ρ ) ∣ ∣<br />
sρ(x kl (t)) (ẋ kl)〉. (1.18)<br />
(g i ◦ s ρ )(x kl ) = i-th column of the matrix s ρ (x kl )<br />
and that s ρ (x kl ) ∈ G ′ is obta<strong>in</strong>ed from the parallel translation along the geodesic for O and<br />
with tangent vector g r+1 (t) <strong>in</strong> O. Hence, for curves x ij , x j i<br />
with i, j ≠ r + 1 we always take<br />
parallel translation along the same geodesic, when we apply s ρ .<br />
When we move g r+1 (t) we consider the parallel translation along different geodesics for each<br />
t. But, as curves x r+1,j , x j r+1 , x 1,r+1 just move the vector g r+1 (t) <strong>in</strong> a real plane generated by<br />
{g r+1 (0), g j (0)} or {g r+1 (0), Jg j (0)}, we have that g 0 (s ρ (x r+1,j )(t)) or g 0 (s ρ (x j r+1 (t))) describes<br />
a circle <strong>in</strong> CK n (ɛ) conta<strong>in</strong>ed <strong>in</strong> the plane generated by {g r+1 (0), g j (0)} (or {g r+1 (0), Jg j (0)}).<br />
From these remarks we have<br />
• (g 0 ◦ s ρ )(x kl ): po<strong>in</strong>t at a distance ρ from O <strong>in</strong> which we arrive along the geodesic with<br />
tangent vector g r+1 at O.<br />
⋄ x r+1,l , l > r + 1.<br />
⋄ x l r+1 , l ≥ r + 1.<br />
(g 0 ◦ s ρ )(x r+1,l ) = cos ɛ (ρ)O + s<strong>in</strong> ɛ (ρ)(cos(t)g r+1 + s<strong>in</strong>(t)g l ).<br />
(g 0 ◦ s ρ )(x l r+1) = cos ɛ (ρ)O + s<strong>in</strong> ɛ (ρ)(cos(t)g r+1 + s<strong>in</strong>(t)(Jg l )).<br />
,
1.5 Space of <strong>complex</strong> r-planes 23<br />
⋄ x 1,r+1 .<br />
(g 0 ◦ s ρ )(x 1,r+1 ) = cos ɛ (ρ)O + s<strong>in</strong> ɛ (ρ)(− s<strong>in</strong>(t)g 1 + cos(t)g r+1 )<br />
s<strong>in</strong>ce g 1 (t) = g 1 cos(t) + g r+1 s<strong>in</strong>(t) and g r+1 (t) is perpendicular to g 1 (t) for all t.<br />
Then, g r+1 (t) = −g 1 s<strong>in</strong>(t) + g r+1 cos(t).<br />
⋄ x r+1<br />
1 .<br />
⋄ x kl , k < l, k, l ≠ r + 1.<br />
(g 0 ◦ s ρ )(x r+1<br />
1 ) = cos ɛ (ρ)O + s<strong>in</strong> ɛ (ρ)(s<strong>in</strong>(t)(Jg 1 ) + cos(t)g r+1 ).<br />
(g 0 ◦ s ρ )(x kl ) = cos ɛ (ρ)O + s<strong>in</strong> ɛ (ρ)g r+1 .<br />
⋄ x l k , k ≤ l, k, l ≠ r + 1. (g 0 ◦ s ρ )(x l k ) = cos ɛ(ρ)O + s<strong>in</strong> ɛ (ρ)g r+1 .<br />
• (g r+1 ◦ s ρ )(x kl ):<br />
⋄ x r+1,l , l > r + 1.<br />
(g r+1 ◦ s ρ )(x r+1,l ) = s −1<br />
ρ (cos(t)g r+1 + s<strong>in</strong>(t)g l )<br />
but this parallel translation co<strong>in</strong>cides with the tangent vector to the geodesic at<br />
time ρ, that is,<br />
s −1<br />
ρ (cos(t)g r+1 +s<strong>in</strong>(t)g l ) = tangent vector to (cos ɛ (ρ)O+s<strong>in</strong> ɛ (ρ)(cos(t)g r+1 +s<strong>in</strong>(t)g l ))<br />
= s<strong>in</strong> ɛ (ρ)O + cos ɛ (ρ)(cos(t)g r+1 + s<strong>in</strong>(t)g l ).<br />
⋄ x l r+1 , l ≥ r + 1.<br />
⋄ x 1,r+1 .<br />
⋄ x r+1<br />
1 .<br />
⋄ x kl , k < l, k, l ≠ j.<br />
(g r+1 ◦ s ρ )(x l r+1) = s<strong>in</strong> ɛ (ρ)O + cos ɛ (ρ)(cos(t)g r+1 + s<strong>in</strong>(t)(Jg l )).<br />
(g r+1 ◦ s ρ )(x 1,r+1 ) = s<strong>in</strong> ɛ (ρ)O + cos ɛ (ρ)(− s<strong>in</strong>(t)g 1 + cos(t)g r+1 ).<br />
(g r+1 ◦ s ρ )(x r+1<br />
1 ) = s<strong>in</strong> ɛ (ρ)O + cos ɛ (ρ)(s<strong>in</strong>(t)(Jg 1 ) + cos(t)g r+1 ).<br />
(g r+1 ◦ s ρ )(x kl ) = s<strong>in</strong> ɛ (ρ)O + cos ɛ (ρ)g r+1 .<br />
⋄ x l k , k ≤ l, k, l ≠ j. (g r+1 ◦ s ρ )(x l k ) = s<strong>in</strong> ɛ(ρ)O + cos ɛ (ρ)g r+1 .<br />
• (g j ◦ s ρ )(x kl ), j > r + 1:<br />
⋄ x jl , l > j.<br />
(g j ◦ s ρ )(x jl ) = s −1<br />
ρ (cos(t)g j + s<strong>in</strong>(t)g l ).
24 Spaces of constant holomorphic curvature<br />
⋄ x l j , l ≥ j.<br />
⋄ x lj , l < j.<br />
⋄ x j l , l ≤ j.<br />
⋄ x kl , k < l, k, l ≠ j.<br />
⋄ x l k<br />
, k ≤ l, k, l ≠ j.<br />
(g<br />
(g j ◦ s ρ )(x l j) = s −1<br />
ρ (cos(t)g j + s<strong>in</strong>(t)(Jg l )).<br />
(g j ◦ s ρ )(x lj ) = s −1<br />
ρ (− s<strong>in</strong>(t)g l + cos(t)g j ).<br />
(g j ◦ s ρ )(x j l ) = s−1 ρ (s<strong>in</strong>(t)(Jg l ) + cos(t)g j ).<br />
(g j ◦ s ρ )(x kl ) = s −1<br />
ρ (g j ).<br />
j ◦ s ρ )(x l k ) = s−1 ρ (g j ).<br />
Now, we compute s ∗ ρα ij (and s ∗ ρβ ij ) <strong>in</strong> terms of α ij , β ij us<strong>in</strong>g (1.18) and evaluat<strong>in</strong>g s ∗ ρα ij<br />
(and s ∗ ρβ ij ) to each curve x kl . Do<strong>in</strong>g so, we obta<strong>in</strong><br />
(s ∗ ρα j1 ) g = −(α j1 ) g ′, j > r + 1.<br />
(s ∗ ρα r+1,1 ) g = − cos ɛ (ρ)(α r+1,1 ) g ′.<br />
(s ∗ ρβ j1 ) g = (β j1 ) g ′, j > r + 1.<br />
(s ∗ ρα r+1,1 ) g = − cos ɛ (ρ)(β r+1,1 ) g ′.<br />
F<strong>in</strong>ally, we have<br />
s ∗ ρ(<br />
∧<br />
j=r+1,...,n<br />
i=1,...,r<br />
α ji ∧ β j1 ) g = cos 2r<br />
ɛ (ρ)(<br />
∧<br />
j=r+1,...,n<br />
i=1,...,r<br />
α j1 ∧ β j1 ) g ′. (1.19)<br />
To get an expression for<br />
i ∗ (<br />
∧<br />
j=r+1,...,n<br />
i=1,...,r<br />
α ji ∧ β j1 )<br />
we use the diagram (1.17) and the computation <strong>in</strong> (1.19), so that<br />
i ∗ (<br />
∧<br />
j=r+1,...,n<br />
i=1,...,r<br />
α ji ∧ β ji ) g = i ∗ ◦ π ∗ 1 ◦ s ∗ ρ(<br />
∧<br />
j=r+1,...,n<br />
i=1,...,r<br />
= i ∗ ◦ π ∗ 1(cos 2r<br />
ɛ (ρ)(<br />
∧<br />
j=r+1,...,n<br />
i=1,...,r<br />
α ji ∧ β ji ) g<br />
α ji ∧ β ji ) ′ g)<br />
= cos 2r<br />
ɛ (ρ)π ∗ 3(dL r[0] ) = cos 2r<br />
ɛ (ρ)dL (n−r)[0]<br />
where we used the expression for the <strong>in</strong>variant density of <strong>complex</strong> r-planes through a po<strong>in</strong>t,<br />
cf. (1.20), and the duality between <strong>complex</strong> r-planes through a po<strong>in</strong>t and the <strong>complex</strong> (n − r)-<br />
planes through the same po<strong>in</strong>t.<br />
Hence, we get<br />
dL r = cos 2r<br />
ɛ (ρ)dx n−r dL (n−r)[O] .
1.5 Space of <strong>complex</strong> r-planes 25<br />
1.5.2 Density of <strong>complex</strong> r-planes conta<strong>in</strong><strong>in</strong>g a fixed <strong>complex</strong> q-plane<br />
We denote by L C r[q]<br />
the <strong>space</strong> of <strong>complex</strong> r-planes conta<strong>in</strong><strong>in</strong>g a fixed <strong>complex</strong> q-plane.<br />
We denote by H [q] := U ɛ (q) the isotropy group of a <strong>complex</strong> q-plane <strong>in</strong> CK n (ɛ) and by H r[q]<br />
the isotropy group of a <strong>complex</strong> r-plane conta<strong>in</strong><strong>in</strong>g the fixed <strong>complex</strong> q-plane. Note that H [q]<br />
acts transitively on L C r[q] .<br />
On the other hand, if we suppose that L 0 q is the fixed <strong>complex</strong> q-plane, then we can def<strong>in</strong>e<br />
the projection<br />
π : H [q] −→ H [q] /H r[q]<br />
g ↦→ L r[q] = gL 0 q.<br />
As the elements <strong>in</strong> H r[q] do not mix tangent vectors to the fixed <strong>complex</strong> q-plane with<br />
orthogonal vectors to this <strong>complex</strong> q-plane, we have<br />
⎧⎛<br />
⎞<br />
⎫<br />
⎨ A 0 0<br />
⎬<br />
H r[q]<br />
∼ = ⎝ 0 B 0 ⎠ , A ∈ U ɛ (q), B ∈ U(r − q), C ∈ U(n − r)<br />
⎩<br />
⎭ ,<br />
0 0 C<br />
π ∗ dL r[q] =<br />
∧<br />
q+1≤i≤r<br />
r+1≤j≤n<br />
α ij ∧ β ij . (1.20)<br />
(The <strong>forms</strong> vanish<strong>in</strong>g, with respect to the isometry group, <strong>in</strong> this case H [q] , are the ones<br />
<strong>in</strong>side the big box.)<br />
1.5.3 Density of <strong>complex</strong> q-planes conta<strong>in</strong>ed <strong>in</strong> a fixed <strong>complex</strong> r-plane<br />
Denote by L (r)<br />
q the <strong>space</strong> of <strong>complex</strong> q-planes conta<strong>in</strong>ed <strong>in</strong> a fixed <strong>complex</strong> r-plane. Let us<br />
fix a <strong>complex</strong> r-plane L r and a <strong>complex</strong> q-plane L (r)<br />
q conta<strong>in</strong>ed <strong>in</strong> L r . Consider the projection<br />
π : U ɛ (r) → U ɛ (r)/U ɛ (q) × U(r − q) where U ɛ (r) denotes the isometry group of the fixed<br />
<strong>complex</strong> r-plane and U ɛ (q) × U(r − q) the isotropy group of a <strong>complex</strong> q-plane conta<strong>in</strong>ed <strong>in</strong><br />
the <strong>complex</strong> r-plane. Then, as <strong>in</strong> the previous case we get<br />
π ∗ dL (r)<br />
q = ∧<br />
1≤i≤q<br />
q+1≤j≤r<br />
α ij ∧ β ij<br />
∧<br />
q+1≤j≤r<br />
α j0 ∧ β j0 =<br />
∧<br />
0≤i≤q<br />
q+1≤j≤r<br />
α ij ∧ β ij .<br />
1.5.4 Measure of <strong>complex</strong> r-planes <strong>in</strong>tersect<strong>in</strong>g a geodesic ball<br />
In order to obta<strong>in</strong> the value of this measure we use the expression for the <strong>in</strong>variant density<br />
of <strong>complex</strong> r-planes <strong>in</strong> (1.16) and the expression of the Jacobian of the map of chang<strong>in</strong>g to<br />
spherical coord<strong>in</strong>ates. This is given by (cf. [Gra73])<br />
cos ɛ (R) s<strong>in</strong>ɛ<br />
2n−1 (R)<br />
|ɛ| (n−1)/2 .<br />
Recall that cos ɛ and s<strong>in</strong> ɛ denote the generalized trigonometric functions def<strong>in</strong>ed at Notation<br />
1.2.2.<br />
We fix as a orig<strong>in</strong> of the spherical coord<strong>in</strong>ates the center of the geodesic ball. Then, us<strong>in</strong>g<br />
the expression <strong>in</strong> spherical coord<strong>in</strong>ates for the element volume of CK n−r (ɛ) (the orthogonal
26 Spaces of constant holomorphic curvature<br />
<strong>space</strong> to a <strong>complex</strong> r-plane <strong>in</strong>tersect<strong>in</strong>g the sphere), we obta<strong>in</strong><br />
∫<br />
∫ ∫<br />
m(L r ∈ L C r : B R ∩ L r ≠ ∅) = dL r =<br />
cos 2r<br />
ɛ (ρ)dx n−r ∧ dG C<br />
B R ∩L r≠∅<br />
G C n,n−r[O]<br />
n,n−r B R ∩L r<br />
∫<br />
= vol(GC n,n−r)<br />
|ɛ| (n−1)/2<br />
cos 2r+1<br />
ɛ (ρ) s<strong>in</strong> 2(n−r)−1<br />
ɛ (ρ)dρdS 2(n−r)−1<br />
∫ R<br />
S 2(n−r)−1 0<br />
= vol(GC n,n−r)<br />
|ɛ| (n−1)/2 vol(S2(n−r)−1 )<br />
= vol(GC n,n−r)<br />
|ɛ| (n−1)/2 vol(S2(n−r)−1 )<br />
= vol(GC n,n−r)<br />
|ɛ| (n−1)/2 vol(S2(n−r)−1 )<br />
= vol(GC n,n−r)<br />
|ɛ| (n−1)/2 vol(S2(n−r)−1 )<br />
∫ R<br />
0<br />
∫ R<br />
0<br />
cos 2r+1<br />
ɛ<br />
(ρ) s<strong>in</strong>ɛ<br />
2(n−r)−1 (ρ)dρ<br />
cos ɛ (ρ)(1 + s<strong>in</strong> 2 ɛ(ρ)) r s<strong>in</strong> 2(n−r)−1<br />
ɛ (ρ)<br />
r∑<br />
( ∫ r R<br />
i)<br />
i=0<br />
r∑<br />
i=0<br />
0<br />
cos ɛ (ρ) s<strong>in</strong>ɛ<br />
2(n−r+i)−1 (ρ)<br />
( 2(n−r+i)<br />
r s<strong>in</strong> ɛ (R)<br />
i)<br />
2(n − r + i) .<br />
At Chapter 4 we give a general expression for the measure of <strong>complex</strong> r-planes <strong>in</strong>tersect<strong>in</strong>g<br />
a regular doma<strong>in</strong>, <strong>in</strong> a way such that the previous expression can be <strong>in</strong>terpreted <strong>in</strong> terms of<br />
mean curvature <strong>in</strong>tegrals and other valuations def<strong>in</strong>ed at Chapter 2.<br />
1.5.5 Reproductive property of Quermass<strong>in</strong>tegrale<br />
At Chapter 3 we prove that the mean curvature <strong>in</strong>tegral does not satisfy a reproductive<br />
property (see Def<strong>in</strong>ition 3.4.1). In this section we prove that Quermass<strong>in</strong>tegrale do satisfy a<br />
reproductive property.<br />
Def<strong>in</strong>ition 1.5.9. Let Ω be a doma<strong>in</strong> <strong>in</strong> CK n (ɛ). For r ∈ {1, . . . , n − 1} we def<strong>in</strong>e<br />
W r (Ω) = (n − r) · O ∫<br />
r−1 · · · O 0<br />
χ(Ω ∩ L r )dL r<br />
n · O n−2 · · · O n−r−1<br />
where O i denotes the area of the sphere of radius <strong>in</strong> the standard Euclidean <strong>space</strong>. Moreover,<br />
we def<strong>in</strong>e<br />
W 0 (Ω) = vol(Ω) and W n (Ω) = O n−1<br />
n<br />
χ(Ω).<br />
Constants are chosen for analogy to the case of real <strong>space</strong> <strong>forms</strong>.<br />
Proposition 1.5.10. Let Ω be a doma<strong>in</strong> <strong>in</strong> CK n (ɛ). Then,<br />
∫<br />
W r (Ω) = c W r (Ω ∩ L q )dL q<br />
for 1 ≤ r ≤ q ≤ n − 1 and c is a constant depend<strong>in</strong>g only on n, r and q.<br />
L C q<br />
Proof. This proof is analogous to the one given by Santaló (cf. [San04]) to obta<strong>in</strong> the result<br />
<strong>in</strong> the Euclidean <strong>space</strong>.<br />
By def<strong>in</strong>ition, it is satisfied<br />
∫<br />
L C q<br />
W r (Ω ∩ L q )dL q = (q − r)O r−1 . . . O 0<br />
qO q−1 . . . O q−r−1<br />
L C r<br />
∫<br />
L C q<br />
∫<br />
Ω∩L r<br />
(q) ≠∅<br />
dL (q)<br />
r ∧ dL q .<br />
We express the densities dL (q)<br />
r , dL q , dL q[r] , dL r <strong>in</strong> terms of the <strong>forms</strong> ϕ ij def<strong>in</strong>ed at Lemma<br />
1.2.7 (we omit the absolute value for the densities),
1.5 Space of <strong>complex</strong> r-planes 27<br />
and<br />
Thus, the equality<br />
dL (q)<br />
r ∧ dL q = ∧<br />
dL q[r] ∧ dL r =<br />
holds.<br />
Apply<strong>in</strong>g it we get the result<br />
∫ ∫<br />
L C q<br />
Ω∩L (q)<br />
r ≠∅<br />
dL q =<br />
∧<br />
q
Chapter 2<br />
Introduction to valuations<br />
The notion of valuation <strong>in</strong> R n was <strong>in</strong>troduced by Blaschke <strong>in</strong> 1955 at [Bla55]. Recently,<br />
Alesker, among others, has extended this notion to differentiable manifolds. A survey about<br />
the development of valuations is given at [MS83] where some references are given.<br />
In this chapter we briefly <strong>in</strong>troduce the theory of valuations, focus<strong>in</strong>g on the concepts and<br />
results we shall use <strong>in</strong> the follow<strong>in</strong>g chapters.<br />
In the last section, we def<strong>in</strong>e some valuations <strong>in</strong> the <strong>space</strong>s of constant holomorphic curvature,<br />
generaliz<strong>in</strong>g the def<strong>in</strong>ition of some valuations <strong>in</strong> C n . We also give relations among the<br />
def<strong>in</strong>ed valuations.<br />
2.1 Def<strong>in</strong>ition and basic properties<br />
Let V be a vector <strong>space</strong> of real dimension n. We denote by K(V ) the set of non-empty compact<br />
convex doma<strong>in</strong>s <strong>in</strong> V .<br />
Def<strong>in</strong>ition 2.1.1. A functional φ : K(V ) → R is called a valuation if<br />
whenever A, B, A ∪ B ∈ K(V ).<br />
φ(A ∪ B) = φ(A) + φ(B) − φ(A ∩ B)<br />
Remark 2.1.2. The extension theorem of Groemer states that every valuation extends uniquely<br />
to the set of f<strong>in</strong>ite union of convex set.<br />
First examples of valuations <strong>in</strong> R n are the volume of a convex doma<strong>in</strong>, the area of its<br />
boundary and its Euler characteristic. Intr<strong>in</strong>sic volumes, def<strong>in</strong>ed by the coefficients of the<br />
polynomial obta<strong>in</strong>ed <strong>in</strong> the Ste<strong>in</strong>er’s formula, are also valuations.<br />
Consider <strong>in</strong> R n a convex doma<strong>in</strong> Ω and denote by Ω r the parallel doma<strong>in</strong> at a distance r<br />
from Ω. Recall that the parallel doma<strong>in</strong> is constituted by all po<strong>in</strong>ts at a distance less or equal<br />
than r.<br />
The Ste<strong>in</strong>er formula relates the volume of the parallel doma<strong>in</strong> with the volume and some<br />
other functionals of the <strong>in</strong>itial doma<strong>in</strong>.<br />
Proposition 2.1.3 (Ste<strong>in</strong>er’s formula). Let Ω ⊂ R n be a compact doma<strong>in</strong> and Ω r the parallel<br />
doma<strong>in</strong> at distance r. Then, the volume of Ω r can be expressed as a polynomial <strong>in</strong> r and<br />
its coefficients are multiples of the valuations V i : K(V ) → R, i ∈ {0, . . . , n}, called <strong>in</strong>tr<strong>in</strong>sic<br />
volumes. The explicit expression is<br />
n∑<br />
vol(Ω r ) = r n−i ω n−i V i (Ω) (2.1)<br />
i=0<br />
where ω n−i denotes the volume of the (n − i)-dimensional ball of radius 1 <strong>in</strong> the Euclidean<br />
<strong>space</strong>.<br />
29
30 Introduction to valuations<br />
Proof. (of the fact that V i are valuations.)<br />
Let A, B, A ∪ B ∈ K(V ). Then, it is satisfied (A ∩ B) r = A r ∩ B r and (A ∪ B) r = A r ∪ B r .<br />
Thus,<br />
vol((A ∪ B) r ) = vol(A r ) + vol(B r ) − vol(A r ∩ B r ) = vol(A r ) + vol(B r ) − vol((A ∩ B) r )<br />
∀r.<br />
Apply<strong>in</strong>g (2.1) we deduce that the <strong>in</strong>tr<strong>in</strong>sic volumes are valuations.<br />
Some particular cases of <strong>in</strong>tr<strong>in</strong>sic volumes are<br />
• V n (Ω) = vol(Ω),<br />
• V n−1 (Ω) = vol(∂Ω)/2,<br />
• V 0 (Ω) = χ(Ω).<br />
Note that the Ste<strong>in</strong>er formula has sense for any convex doma<strong>in</strong>, without any assumption<br />
on the regularity of the boundary. In some cases it is <strong>in</strong>terest<strong>in</strong>g to consider convex doma<strong>in</strong>s<br />
such that its boundary is an oriented hypersurface of class C 2 . Apply<strong>in</strong>g the previous formula<br />
<strong>in</strong> this case we obta<strong>in</strong> the so-called mean curvature <strong>in</strong>tegrals.<br />
Def<strong>in</strong>ition 2.1.4. Let S be a hypersurface of class C 2 <strong>in</strong> a Riemannian manifold M of dimension<br />
n. If x ∈ S, we denote the second fundamental form of S at x by II x . We def<strong>in</strong>e the i-th<br />
mean curvature <strong>in</strong>tegral of S as<br />
( ) n − 1 −1 ∫<br />
M i (S) =<br />
i<br />
S<br />
σ i (II x )dx<br />
where σ i (II x ) denotes the r-th symmetric elementary function of the second fundamental form<br />
II x .<br />
Then, the Ste<strong>in</strong>er formula <strong>in</strong> R n is<br />
n−1<br />
(<br />
∑ n<br />
)<br />
vol(Ω r ) = vol(Ω) + r n−i i<br />
n M n−i−1(∂Ω).<br />
i=0<br />
Sometimes, it is def<strong>in</strong>ed M −1 (∂Ω) := vol(Ω).<br />
The relation among the <strong>in</strong>tr<strong>in</strong>sic volumes and the mean curvature <strong>in</strong>tegrals, for convex<br />
doma<strong>in</strong>s with boundary of class C 2 , is<br />
( n<br />
i)<br />
V i (Ω) = M n−i−1 (∂Ω).<br />
nω n−i<br />
Def<strong>in</strong>ition 2.1.5. A valuation φ is cont<strong>in</strong>uous if it is cont<strong>in</strong>uous with respect to the Hausdorff<br />
metric.<br />
Recall that the Hausdorff distance between two compact sets A, B is given by<br />
d Haus (A, B) = max{sup <strong>in</strong>f {d(a, b)}, sup <strong>in</strong>f {d(a, b)}}<br />
a∈A b∈B<br />
b∈B a∈A<br />
where d(a, b) is the distance def<strong>in</strong>ed <strong>in</strong> the ambient <strong>space</strong> for A and B.<br />
Example 2.1.6. The <strong>in</strong>tr<strong>in</strong>sic volumes are cont<strong>in</strong>uous valuations. Anyway, there are some<br />
<strong>in</strong>terest<strong>in</strong>g examples of non-cont<strong>in</strong>uous valuations <strong>in</strong> R n . For example, the aff<strong>in</strong>e surface area<br />
is a valuation <strong>in</strong> the Euclidean <strong>space</strong>, but it is not cont<strong>in</strong>uous (cf. [KR97]). The aff<strong>in</strong>e surface<br />
area of a convex doma<strong>in</strong> Ω ⊂ R n is def<strong>in</strong>ed as the <strong>in</strong>tegral of the (n + 1)-th root of the
2.1 Def<strong>in</strong>ition and basic properties 31<br />
generalized Gauss curvature (cf. [Sch93] notes of Sections 1.5 and 2.5) of the boundary of the<br />
doma<strong>in</strong> with respect to the (n − 1)-dimensional Hausdorff measure of the boundary<br />
∫<br />
n+1<br />
AS(Ω) =<br />
√ K x dx.<br />
∂Ω<br />
One of the most important property of this valuation is that it is <strong>in</strong>variant under translations<br />
and l<strong>in</strong>ear transformations with determ<strong>in</strong>ant 1.<br />
Def<strong>in</strong>ition 2.1.7. Given a (2n − 1)-form ω def<strong>in</strong>ed on S(V ), and a smooth measure, η we<br />
consider, for each regular doma<strong>in</strong> Ω,<br />
∫ ∫<br />
η + ω<br />
N(Ω)<br />
Ω<br />
where N(Ω) denotes the normal bundle of the boundary of the doma<strong>in</strong>. The obta<strong>in</strong>ed functional<br />
is called smooth valuation.<br />
Def<strong>in</strong>ition 2.1.8. Let Ω ∈ K(V ). A valuation φ : K(V ) → R is called<br />
• translation <strong>in</strong>variant if<br />
φ(ψΩ) = φ(Ω)<br />
for every ψ translation of the vector <strong>space</strong> V ;<br />
• <strong>in</strong>variant with respect to a group G act<strong>in</strong>g on V if<br />
for every g ∈ G;<br />
• homogeneous of degree k if<br />
φ(gΩ) = φ(Ω)<br />
φ(λΩ) = λ k φ(Ω) for every λ > 0, k ∈ R;<br />
• even (resp. odd) if<br />
with ɛ even (resp. odd);<br />
φ(−1 · Ω) = (−1) ɛ φ(Ω)<br />
• monotone if<br />
φ(Ω 1 ) ≥ φ(Ω 2 ) for every Ω 1 , Ω 2 ∈ K(V ) and Ω 1 ⊃ Ω 2 .<br />
The <strong>space</strong> of cont<strong>in</strong>uous <strong>in</strong>variant translation valuations is denoted by Val(V ), the sub<strong>space</strong><br />
of Val(V ) of the homogeneous valuations of degree k by Val k (V ) and the sub<strong>space</strong> of Val(V )<br />
of even valuations (resp. odd valuations) by Val + (V ) (resp. Val − (V )).<br />
Example 2.1.9. The <strong>in</strong>tr<strong>in</strong>sic volume V i is a cont<strong>in</strong>uous <strong>in</strong>variant translation valuation homogeneous<br />
of degree i.<br />
Remark 2.1.10. The <strong>space</strong> Val(V ) has structure of <strong>in</strong>f<strong>in</strong>ite dimensional vector <strong>space</strong>.<br />
The follow<strong>in</strong>g result of P. McMullen [McM77] gives a decomposition of the <strong>space</strong> of valuations<br />
depend<strong>in</strong>g on the degree, and another depend<strong>in</strong>g on the parity<br />
Theorem 2.1.11 ([McM77]). Let n = dim V . Then,<br />
Val(V ) =<br />
n⊕<br />
Val i (V ) and Val(V ) = Val + (V ) ⊕ Val − (V ).<br />
i=0
32 Introduction to valuations<br />
The l<strong>in</strong>ear group GL(V ) of <strong>in</strong>vertible l<strong>in</strong>ear transformations of V acts transitively on Val(V )<br />
(gφ)(Ω) = φ(g −1 (Ω)) for g ∈ GL(V ), φ ∈ Val(V ), Ω ∈ K(V ).<br />
This action is cont<strong>in</strong>uous and preserves the homogeneous degree of the valuation.<br />
Theorem 2.1.12 (Irreducibility Theorem). Let V be an n-dimensional vector <strong>space</strong>. The<br />
natural representation of GL(V ) at Val + i (V ) and Val− i<br />
(V ) is irreducible for any i ∈ {0, . . . , n}.<br />
That is, there is no proper closed GL(V )-<strong>in</strong>variant sub<strong>space</strong>.<br />
From this theorem, it can be proved the follow<strong>in</strong>g result which relates cont<strong>in</strong>uous valuations<br />
with smooth valuations.<br />
Theorem 2.1.13 ([Ale01]). In a vector <strong>space</strong> V , the smooth translation <strong>in</strong>variant valuations<br />
are dense <strong>in</strong> the <strong>space</strong> of cont<strong>in</strong>uous translation <strong>in</strong>variant valuations.<br />
If V has an Euclidean metric, then every group G subgroup of the orthogonal group, acts on<br />
Val(V ) and it has sense to consider the <strong>space</strong> Val G (V ) ⊂ Val(V ), i.e. the <strong>space</strong> of G-<strong>in</strong>variant<br />
valuations under the action of the group G ⋉ V . The follow<strong>in</strong>g result by Alesker gives the<br />
necessary and sufficient condition to be this <strong>space</strong> of f<strong>in</strong>ite dimension.<br />
Corollary 2.1.14 ([Ale07a] Proposition 2.6). The <strong>space</strong> Val G (V ) is f<strong>in</strong>ite dimensional if and<br />
only of G acts transitively on the unit sphere of V .<br />
2.2 Hadwiger Theorem<br />
In 1957 Hadwiger proved the follow<strong>in</strong>g result concern<strong>in</strong>g valuations.<br />
Theorem 2.2.1 ([Had57]). The dimension of the <strong>space</strong> of cont<strong>in</strong>uous translation and O(n)-<br />
<strong>in</strong>variant valuations is<br />
dim Val O(n) (R n ) = n + 1<br />
and a basis of this <strong>space</strong> is given by<br />
where V i denotes the i-th <strong>in</strong>tr<strong>in</strong>sic volume.<br />
V 0 , V 1 , . . . , V n−1 , V n<br />
Remark 2.2.2. From this theorem it follows that <strong>in</strong> R n the sub<strong>space</strong> of valuations of homogeneous<br />
degree k ∈ {0, . . . , n} is of dimension 1.<br />
Last remark allows us to prove, <strong>in</strong> an easy way, some of the classical results of <strong>in</strong>tegral<br />
<strong>geometry</strong> (<strong>in</strong> R n ), such as reproductive or the k<strong>in</strong>ematic formulas.<br />
Example 2.2.3. • Crofton formula. Let Ω ⊂ R n be a compact convex doma<strong>in</strong>, and let L r<br />
be the <strong>space</strong> of all planes of dimension r with dL r the (unique up to a constant factor)<br />
<strong>in</strong>variant density. The measure of the set of planes a convex body <strong>in</strong> R n can be expressed<br />
<strong>in</strong> terms of the <strong>in</strong>tr<strong>in</strong>sic volumes as<br />
∫<br />
χ(Ω ∩ L r )dL r = cV n−r (Ω).<br />
L r<br />
This expression is obta<strong>in</strong>ed from the fact that the <strong>in</strong>tegral <strong>in</strong> the left hand side is a<br />
valuation of homogeneous degree (n − r).<br />
At Chapter 4, we study the expression of the measure of <strong>complex</strong> planes meet<strong>in</strong>g a<br />
doma<strong>in</strong> <strong>in</strong> CK n (ɛ), and at Chapter 5, the measure of totally real planes of dimension n<br />
meet<strong>in</strong>g a doma<strong>in</strong>.
2.2 Hadwiger Theorem 33<br />
• Reproductive property of the <strong>in</strong>tr<strong>in</strong>sic volumes. If Ω is a compact convex doma<strong>in</strong>, the<br />
reproductive formula for the <strong>in</strong>tr<strong>in</strong>sic volumes <strong>in</strong> R n is given by<br />
∫<br />
V (r)<br />
i<br />
L r<br />
(∂Ω ∩ L r )dL r = cV i (∂Ω)<br />
where L r denotes the <strong>space</strong> of r-planes <strong>in</strong> R n and c is a constant depend<strong>in</strong>g only on n,<br />
r and i.<br />
Clearly, the <strong>in</strong>tegral <strong>in</strong> the left hand side is a cont<strong>in</strong>uous valuation of homogeneous<br />
degree i. By the Hadwiger Theorem the equality follows directly, s<strong>in</strong>ce the i-th <strong>in</strong>tr<strong>in</strong>sic<br />
volume is a homogeneous valuation of this degree and the dimension of the <strong>space</strong> of<br />
homogeneous valuations of degree i is 1. In order to compute the value of c it is used<br />
the so-called template method, which consists on evaluat<strong>in</strong>g each side of the equation <strong>in</strong><br />
an easy doma<strong>in</strong> (for <strong>in</strong>stance, a sphere) and then compute the value of c. In [San04] it is<br />
given a way to prove this reproductive formula, but us<strong>in</strong>g the mean curvature <strong>in</strong>tegrals.<br />
The value of c it is also computed.<br />
• K<strong>in</strong>ematic formula. Although <strong>in</strong> this work we do not study k<strong>in</strong>ematic formulas, we would<br />
like to give, for completion, the classical k<strong>in</strong>ematic formula of Blaschke-Santaló.<br />
One of the problems of study of the <strong>in</strong>tegral <strong>geometry</strong> consists on measur<strong>in</strong>g the movements<br />
of R n which takes one convex doma<strong>in</strong> to another fixed one. In R n , if we denote<br />
by O(n) the movements group of the <strong>space</strong>, we obta<strong>in</strong> the classical k<strong>in</strong>ematic formula<br />
∫<br />
O(n)<br />
χ(Ω 1 ∩ gΩ 2 )dg =<br />
n∑<br />
c n,i V i (Ω 1 )V n−i (Ω 2 ). (2.2)<br />
This formula can be proved from apply<strong>in</strong>g twice the Hadwiger Theorem, and then,<br />
apply<strong>in</strong>g the obta<strong>in</strong>ed formula to spheres of different radius.<br />
Note that the <strong>in</strong>tegral on the left hand side is a functional on the first convex doma<strong>in</strong><br />
Ω 1 , but also on the second convex doma<strong>in</strong> Ω 2 . As a functional on the second convex<br />
doma<strong>in</strong>, from the Hadwiger Theorem, we have<br />
∫<br />
n∑<br />
χ(Ω 1 ∩ gΩ 2 )dg = c i (Ω 1 )V i (Ω 2 )<br />
O(n)<br />
where the coefficients c i (Ω 1 ) depend on Ω 1 . But, the <strong>in</strong>tegral under consideration is also<br />
a valuation with respect to Ω 1 , thus, the coefficients c i (Ω 1 ) are valuations and, aga<strong>in</strong> by<br />
Hadwiger Theorem, we obta<strong>in</strong> that it is satisfied<br />
∫<br />
n∑ n∑<br />
χ(Ω 1 ∩ gΩ 2 )dg = c ij V j (Ω 1 )V i (Ω 2 ).<br />
O(n)<br />
i=0<br />
i=0<br />
i=0 j=0<br />
To obta<strong>in</strong> the expression (2.2), first, note that the desired expression have to be symmetric<br />
with respect to Ω 1 and Ω 2 , hence, c ij = c ji . To prove that most of the constants<br />
c ij vanishes we use the template method, i.e. we apply the equality for a sphere of radius<br />
r and for a sphere of radius R.<br />
Us<strong>in</strong>g the <strong>in</strong>variance with respect to the rotations of a sphere we have<br />
∫<br />
∫ ∫<br />
χ(B r ∩ gB R )dg = χ(B r ∩ (φB R + v))dvdφ (2.3)<br />
O(n)<br />
O(n) R n ∫<br />
= vol(O(n)) χ(B r ∩ (B R + v))dv = vol(O(n))(r + R) n ω n .<br />
R n
34 Introduction to valuations<br />
On the other hand, as the <strong>in</strong>tr<strong>in</strong>sic volume V i is a homogeneous valuation of degree i we<br />
get<br />
n∑ n∑<br />
n∑<br />
c ij V j (B r )V i (B R ) = c ij r j R i V j (B 1 )V i (B 1 ). (2.4)<br />
i=0 j=0<br />
Thus, <strong>in</strong> both cases (2.3) and (2.4), we obta<strong>in</strong>ed a polynomial on r and R. Compar<strong>in</strong>g<br />
the coefficients we get i + j = n <strong>in</strong> the last summation <strong>in</strong> (2.4).<br />
To get the explicit value for the constant, it rema<strong>in</strong>s only to use the expression of the<br />
<strong>in</strong>tr<strong>in</strong>sic volume of sphere of any radius. As the parallel doma<strong>in</strong> at distance r of a sphere<br />
of radius R is a sphere of radius r + R, we can easily compute its <strong>in</strong>tr<strong>in</strong>sic volume us<strong>in</strong>g<br />
the Ste<strong>in</strong>er formula, and we get<br />
2.3 Alesker Theorem<br />
i,j=0<br />
V i (B r ) = r i ( n<br />
i<br />
)<br />
ωn<br />
ω n−i<br />
.<br />
Recently, Alesker gave an analogous theorem of Hadwiger Theorem for the Hermitian standard<br />
<strong>space</strong> V = C n , with isometry group IU(n) = C n ⋊U(n). As the isometry group of C n is smaller<br />
than the isometry group of R 2n it may happen that some non-<strong>in</strong>variant valuations under the<br />
isometry group of R 2n is <strong>in</strong>variant under the isometry group of C n , and this occurs.<br />
Theorem 2.3.1 ([Ale03] Theorem 2.1.1). Let Val U(n) (C n ) be the <strong>space</strong> of cont<strong>in</strong>uous translation<br />
and U(n)-<strong>in</strong>variant valuations <strong>in</strong> C n . Then,<br />
( ) n + 2<br />
dim Val U(n) (C n ) = ,<br />
2<br />
and the dimension of the sub<strong>space</strong> of degree k homogeneous valuations is<br />
m<strong>in</strong>{k, 2n − k}<br />
2<br />
+ 1.<br />
Alesker [Ale03] also gave two bases of cont<strong>in</strong>uous isometry <strong>in</strong>variant valuations on C n . One<br />
of these bases is def<strong>in</strong>ed as the <strong>in</strong>tegral of the projection volume. That is, let Ω ⊂ C n be a<br />
convex doma<strong>in</strong> and k, l <strong>in</strong>tegers such that 0 ≤ k ≤ 2l ≤ 2n, then<br />
∫<br />
C k,l (Ω) :=<br />
G C n,l<br />
V k (Pr Ll (Ω))dL l<br />
where G C n,l denotes the <strong>space</strong> of <strong>complex</strong> l-planes <strong>in</strong> Cn through the orig<strong>in</strong> (see Section 1.5),<br />
Pr Ll (Ω) denotes the orthogonal projection of Ω at L l and V k the k-th <strong>in</strong>tr<strong>in</strong>sic volume. Valuations<br />
{C k,l } with 0 ≤ k ≤ 2l ≤ 2n def<strong>in</strong>e a basis of Val U(n) (C n ). The <strong>in</strong>dex k co<strong>in</strong>cides with<br />
the homogeneous degree of the valuation.<br />
The other basis of Val U(n) (C n ) is {U k,p } with k, p <strong>in</strong>tegers such that 0 ≤ 2p ≤ k ≤ 2n with<br />
∫<br />
U k,p (Ω) :=<br />
L C n−p<br />
V k−2p (Ω ∩ L n−p )dL n−p (2.5)<br />
where L C n−p denotes the <strong>space</strong> of <strong>complex</strong> aff<strong>in</strong>e (n − p)-planes of C n (see Section 1.5), and<br />
V k−2p the (k − 2p)-th <strong>in</strong>tr<strong>in</strong>sic volume. The <strong>in</strong>dex k co<strong>in</strong>cides with the homogeneous degree of<br />
the valuation.
2.3 Alesker Theorem 35<br />
Example 2.3.2. We describe explicitily the elements of {U k,p } <strong>in</strong> C 3 . In C 3 there are ( 5<br />
2)<br />
= 10<br />
l<strong>in</strong>early <strong>in</strong>dependent valuations. For each degree k we have as much valuations as l<strong>in</strong>early<br />
<strong>in</strong>dependent <strong>in</strong>tegers p such that<br />
0 ≤ p ≤<br />
m<strong>in</strong>{k, 2n − k}<br />
.<br />
2<br />
Suppose that Ω is a convex doma<strong>in</strong> <strong>in</strong> C 3 . Then, a basis of Val U(3) (C 3 ) is given by the follow<strong>in</strong>g<br />
valuations.<br />
k = 0: There is only one value of p, p = 0 and<br />
k = 1: There is only one value of p, p = 0 and<br />
k = 2: There are two values of p, p = 0, 1.<br />
If p = 0 then<br />
U 0,0 (Ω) = V 0 (Ω) = χ(Ω).<br />
U 1,0 (Ω) = V 1 (Ω).<br />
U 2,0 (Ω) = V 2 (Ω).<br />
If p = 1 then<br />
∫<br />
U 2,1 (Ω) = V 0 (Ω ∩ L 2 )dL 2 .<br />
L 2<br />
k = 3: There are two values of p, p = 0, 1.<br />
If p = 0 then<br />
U 3,0 (Ω) = V 3 (Ω).<br />
If p = 1 then<br />
∫<br />
U 3,1 (Ω) = V 1 (Ω ∩ L 2 )dL 2 .<br />
L 2<br />
k = 4: There are two values of p, p = 0, 1.<br />
If p = 0 then<br />
U 4,0 (Ω) = V 4 (Ω).<br />
If p = 1 then<br />
∫<br />
U 4,1 (Ω) = V 2 (Ω ∩ L 2 )dL 2 .<br />
L 2<br />
k = 5: There is only one value of p, p = 0 and<br />
U 5,0 (Ω) = V 5 (Ω) = 1 2 vol(∂Ω).<br />
k = 6: There is only one value of p, p = 0 and<br />
U 6,0 (Ω) = V 6 (Ω) = vol(Ω).<br />
Note that we obta<strong>in</strong>ed all <strong>in</strong>tr<strong>in</strong>sic volumes V j (K), j ∈ {0, . . . , 6}, <strong>in</strong> the same way as <strong>in</strong><br />
R 6 , but at C 3 appear three new l<strong>in</strong>ear <strong>in</strong>dependent valuations.
36 Introduction to valuations<br />
In the same way as <strong>in</strong> R n , hav<strong>in</strong>g a basis for Val U(n) (C n ) allows us to establish a k<strong>in</strong>ematic<br />
formula. The difference is that, now, does not seem possible to f<strong>in</strong>d the constants us<strong>in</strong>g the<br />
template method. Alesker, <strong>in</strong> the same paper [Ale03] establish the follow<strong>in</strong>g result.<br />
Theorem 2.3.3 ([Ale03] Theorem 3.1.1). Let Ω 1 , Ω 2 ⊂ C n be doma<strong>in</strong>s with piecewise smooth<br />
boundaries such that for any U ∈ IU(n) the <strong>in</strong>tersection Ω 1 ∩ U(Ω 2 ) has a f<strong>in</strong>ite number of<br />
components. Then,<br />
∫<br />
χ(Ω 1 ∩ U(Ω 2 ))dU =<br />
∑ ∑<br />
κ(k 1 , k 2 , p 1 , p 2 )U k1 ,p 1<br />
(Ω 1 )U k2 ,p 2<br />
(Ω 2 ),<br />
U∈IU(n)<br />
p 1 ,p 2<br />
k 1 +k 2 =2n<br />
where the <strong>in</strong>dex of the <strong>in</strong>ner summation runs over 0 ≤ p i ≤ k i /2, i = 1, 2, and κ(k 1 , k 2 , p 1 , p 2 )<br />
are constants depend<strong>in</strong>g only on n, k 1 , k 2 , p 1 , p 2 .<br />
Theorem 2.3.4 ([Ale03] Theorem 3.1.2). Let Ω ⊂ C n be a doma<strong>in</strong> with piecewise smooth<br />
boundary and 0 < q < n, 0 < 2p < k < 2q. Then,<br />
∫<br />
L C r<br />
U k,p (Ω ∩ L r )dL r =<br />
[k/2]+n−q<br />
∑<br />
p=0<br />
where γ p are constants depend<strong>in</strong>g only on n, q, and p.<br />
γ p · U k+2(n−q),p (Ω),<br />
Theorem 2.3.5 ([Ale03] Theorem 3.1.3). Let Ω ⊂ C n be a doma<strong>in</strong> with piecewise smooth<br />
boundary. Then,<br />
∫<br />
[n/2]<br />
∑<br />
χ(Ω ∩ L n )dL n = β p · U n,p (Ω),<br />
L R n<br />
where L R n denotes the <strong>space</strong> of Lagrangian planes (i.e. totally real planes of dimension n) <strong>in</strong><br />
C n and β p are constants depend<strong>in</strong>g only on n and p.<br />
The constants for Theorem 2.3.3 are given by Bernig-Fu at [BF08]. These constants were<br />
computed us<strong>in</strong>g <strong>in</strong>direct methods and others bases of valuations <strong>in</strong> C n . In this work, we give<br />
the constant, <strong>in</strong> some cases, for Theorems 2.3.4 and 2.3.5.<br />
In [BF08] are given some other bases for Val U(n) (C n ). In Chapters 4 and 5, we use a basis<br />
def<strong>in</strong>ed <strong>in</strong> [BF08] and we extend it to all <strong>complex</strong> <strong>space</strong> <strong>forms</strong> CK n (ɛ). The def<strong>in</strong>ition of theses<br />
valuations and its extension <strong>in</strong> CK n (ɛ) is given at Section 2.4.2.<br />
F<strong>in</strong>ally, we note that Proposition 2.1.14 gives another way to generalize the theory of valuations<br />
<strong>in</strong> vector <strong>space</strong>s. From this proposition, we have that for any group act<strong>in</strong>g transitively<br />
on the sphere can be stated a Hadwiger type theorem, i.e. the <strong>space</strong> of cont<strong>in</strong>uous translation<br />
<strong>in</strong>variant valuations has f<strong>in</strong>ite dimension, thus, it has sense to compute its dimension and give<br />
a basis. An expression for the k<strong>in</strong>ematic formula can also be given.<br />
In this section, we recalled the case <strong>in</strong> which the act<strong>in</strong>g group is U(n) and befors we studied<br />
SO(n), but there are some known result for other groups.<br />
The groups act<strong>in</strong>g over a sphere are classified and they are (cf. [Bor49], [Bor50], [MS43])<br />
six <strong>in</strong>f<strong>in</strong>ite series<br />
and three exceptional groups<br />
p=0<br />
SO(n), U(n), SU(n), Sp(n), Sp(n) · U(1), Sp(n) · Sp(1)<br />
G 2 , Sp<strong>in</strong>(7), Sp<strong>in</strong>(9).<br />
Alesker and Bernig obta<strong>in</strong>ed a Hadwiger type theorem, a k<strong>in</strong>ematic formula (and the<br />
algebraic structure) for some of these groups (cf. [Ale04] for G = SU(2), [Ber08a] for G =<br />
SU(n) and [Ber08b] for G = G 2 and G = Sp<strong>in</strong>(7)).
2.4 Valuations on <strong>complex</strong> <strong>space</strong> <strong>forms</strong> 37<br />
2.4 Valuations on <strong>complex</strong> <strong>space</strong> <strong>forms</strong><br />
In this section we def<strong>in</strong>e some valuations on CK n (ɛ) and we give relations among them.<br />
2.4.1 Smooth valuations on manifolds<br />
The notion of smooth valuation <strong>in</strong> a differentiable manifold was recently studied (cf. [Ale06a],<br />
[Ale06b], [AF08], [Ale07b], [AB08]). First, a def<strong>in</strong>ition of smooth valuation on a manifold was<br />
given, and then it was proved that some important properties, which we do not study <strong>in</strong> this<br />
work, of smooth valuations, such as the duality property, still hold.<br />
Def<strong>in</strong>ition 2.1.7 can be extended for differentiable manifolds.<br />
Def<strong>in</strong>ition 2.4.1. Let M be a differentiable manifold and Ω a compact submanifold. Given<br />
a (2n − 1)-form ω <strong>in</strong> S(M), and a smooth measure η, we consider for any Ω<br />
∫ ∫<br />
η + ω,<br />
Ω<br />
where N(Ω) denotes the normal cycle (cf. [Ale07a]). The obta<strong>in</strong>ed functional is called smooth<br />
valuation.<br />
Remark 2.4.2. A more general def<strong>in</strong>ition analogue to Def<strong>in</strong>iton 2.1.1 appears <strong>in</strong> [Ale06b], and<br />
it is called f<strong>in</strong>itely additive measure.<br />
In <strong>space</strong>s of constant sectional curvatures, <strong>in</strong>variant smooth valuations are well-known.<br />
These <strong>space</strong>s have the same isotropy group of a po<strong>in</strong>t as a po<strong>in</strong>t <strong>in</strong> R n , and from the po<strong>in</strong>t of<br />
view of a homogeneous <strong>space</strong>s they can be studied <strong>in</strong> an analogous way. Despite this fact, a<br />
Hadwiger type theorem for cont<strong>in</strong>uous valuations (and not only for smooth valuations) it is<br />
not known, i.e. it is not known a basis of cont<strong>in</strong>uous translation <strong>in</strong>variant valuations <strong>in</strong>variant<br />
also for the isometry group of the <strong>space</strong>. The dimension of this <strong>space</strong> of valuations it is not<br />
known an analogous result to Theorem 2.1.13.<br />
Anyway, a big amount of the results <strong>in</strong> <strong>in</strong>tegral <strong>geometry</strong> are known <strong>in</strong> these <strong>space</strong>s. For<br />
<strong>in</strong>stance, Santaló [San04, page 309] proved that a reproductive formula holds for any real<br />
<strong>space</strong> form and also obta<strong>in</strong>ed an expression for the measure of totally geodesic planes meet<strong>in</strong>g<br />
a convex doma<strong>in</strong>.<br />
In view of these results of Santaló and the knowledge of a basis of cont<strong>in</strong>uous <strong>in</strong>variant<br />
valuations on C n , the aim of this work is to study the classical formulas <strong>in</strong> <strong>in</strong>tegral <strong>geometry</strong><br />
described <strong>in</strong> the last paragraph <strong>in</strong> <strong>complex</strong> <strong>space</strong> <strong>forms</strong>, i.e. <strong>in</strong> the standard Hermitian <strong>space</strong>,<br />
and <strong>in</strong> the <strong>complex</strong> projective and hyperbolic <strong>space</strong>.<br />
2.4.2 Hermitian <strong>in</strong>tr<strong>in</strong>sic volumes<br />
Bernig and Fu [BF08] def<strong>in</strong>ed the Hermitian <strong>in</strong>tr<strong>in</strong>sic volumes <strong>in</strong> C n . In this section we recall<br />
this def<strong>in</strong>ition and its extension to CK n (ɛ) follow<strong>in</strong>g [Par02].<br />
Bernig and Fu at [BF08, page 14] def<strong>in</strong>ed <strong>in</strong> T C n , the follow<strong>in</strong>g <strong>in</strong>variant 1-<strong>forms</strong> α, β<br />
and γ and the <strong>in</strong>variant 2-<strong>forms</strong> θ 0 , θ 1 , θ 2 and θ s . Let (z 1 , . . . , z n , ζ 1 , . . . , ζ n ) be the canonical<br />
coord<strong>in</strong>ates of T C n ≃ C n × C n with z i = x i + √ −1y i and ζ i = ξ i + √ −1η i . Then,<br />
N(Ω)<br />
θ 0 := ∑ n<br />
i=1 dξ i ∧ dη i , θ 1 := ∑ n<br />
i=1 (dx i ∧ dη i − dy i ∧ dξ i ) ,<br />
θ 2 := ∑ n<br />
i=1 dx i ∧ dy i , θ s := ∑ n<br />
i=1 (dx i ∧ dξ i + dy i ∧ dη i ) ,<br />
α := ∑ n<br />
i=1 ξ idx i + η i dy i , β := ∑ n<br />
i=1 ξ idy i − η i dx i ,<br />
γ := ∑ n<br />
i=1 ξ idη i − η i dξ i .
38 Introduction to valuations<br />
α is the contact form (see Remark 1.3.8) and θ s is the symplectic form of T C n . Recall that a 2-<br />
form ω def<strong>in</strong>ed <strong>in</strong> a manifold of dimension 2m is said symplectic if it is closed, non-degenerated<br />
and ω m ≠ 0.<br />
The previous <strong>forms</strong> are only def<strong>in</strong>ed for ɛ = 0 s<strong>in</strong>ce canonical coord<strong>in</strong>ates exist only at C n .<br />
But, we can express them <strong>in</strong> terms of a mov<strong>in</strong>g frame on S(C n ) (and, then extend them for<br />
any ɛ ∈ R). The expression <strong>in</strong> terms of a mov<strong>in</strong>g frame allows us to prove that these <strong>forms</strong> are<br />
well-def<strong>in</strong>ed, i.e. they do not depend on the chosen coord<strong>in</strong>ates. Let (x, v) ∈ S(CK n (ɛ)) and<br />
let {x; e 1 := v, Je 1 , . . . , e n , Je n } be a J-mov<strong>in</strong>g frame def<strong>in</strong>ed <strong>in</strong> a neighborhood of x. Then<br />
θ 0 =<br />
θ 1 =<br />
θ 2 =<br />
n∑<br />
α 1i ∧ β 1i ,<br />
i=2<br />
n∑<br />
(α i ∧ β 1i − β i ∧ α 1i ), (2.6)<br />
i=2<br />
n∑<br />
α i ∧ β i ,<br />
i=2<br />
where α i , β i , α ij , β ij are the <strong>forms</strong> given at (1.12) but <strong>in</strong>terpreted as <strong>forms</strong> <strong>in</strong> S(CK n (ɛ)).<br />
Remark 2.4.3. From the expression of θ 0 , θ 1 and θ 2 <strong>in</strong> terms of a mov<strong>in</strong>g frame we can def<strong>in</strong>e<br />
these 2-<strong>forms</strong> <strong>in</strong> S(CK n (ɛ)) for any ɛ.<br />
Remark 2.4.4. In [Par02] <strong>in</strong>variant 2-<strong>forms</strong> at CK n (ɛ) are def<strong>in</strong>ed <strong>in</strong> the same way as before<br />
(see Section 2.4.3).<br />
Proposition 2.4.5 ([Par02] Proposition 2.2.1). The algebra Ω ∗ (S(CK n (ɛ))) of R-valued <strong>in</strong>variant<br />
differential <strong>forms</strong> on the unit tangent bundle S(CK n (ɛ)) is generated by<br />
α, β, γ, θ 0 , θ 1 , θ 2 , θ s .<br />
From this proposition, Bernig and Fu def<strong>in</strong>e the families of (2n−1)-<strong>forms</strong> {β k,q } and {γ k,q }<br />
at S(C n ), but from the def<strong>in</strong>ition of θ 0 , θ 1 and θ 2 at S(CK n (ɛ)) these families of <strong>forms</strong> can be<br />
def<strong>in</strong>ed <strong>in</strong> the same way at S(CK n (ɛ)). By the previous proposition we have, as it is note <strong>in</strong><br />
[Par02], that all (2n − 1)-form <strong>in</strong>variant on S(CK n (ɛ)) such that they do not vanish over the<br />
normal bundle of a doma<strong>in</strong> Ω (cf. Lemma 1.3.12) are the ones given <strong>in</strong> the next def<strong>in</strong>ition.<br />
Def<strong>in</strong>ition 2.4.6. Let k, q ∈ N be such that max{0, k − n} ≤ q ≤ k 2<br />
differential (2n − 1)-<strong>forms</strong> at S(CK n (ɛ)) are def<strong>in</strong>ed<br />
< n. Then, the follow<strong>in</strong>g<br />
where<br />
β k,q := c n,k,q β ∧ θ n−k+q<br />
0 ∧ θ k−2q−1<br />
1 ∧ θ q 2 , k ≠ 2q<br />
γ k,q := c n,k,q<br />
2 γ ∧ θn−k+q−1 0 ∧ θ k−2q<br />
1 ∧ θ q 2 , n ≠ k − q<br />
1<br />
c n,k,q :=<br />
q!(n − k + q)!(k − 2q)!ω 2n−k<br />
and ω 2n−k denotes the volume of the Euclidean ball of radius 1 and dimension 2n − k.<br />
Def<strong>in</strong>ition 2.4.7. Given a regular doma<strong>in</strong> Ω ⊂ CK n (ɛ), <strong>forms</strong> β k,q and γ k,q def<strong>in</strong>e the follow<strong>in</strong>g<br />
<strong>in</strong>variant valuations (see Section 2.4.3) <strong>in</strong> CK n (ɛ) (for max{0, k − n} ≤ q ≤ k 2 < n)<br />
∫<br />
∫<br />
B k,q (Ω) := β k,q (k ≠ 2q) and Γ k,q (Ω) = γ k,q (n ≠ k − q)<br />
N(Ω)<br />
N(Ω)<br />
where N(Ω) denotes the normal fiber bundle of Ω (see Def<strong>in</strong>ition 1.3.10).
2.4 Valuations on <strong>complex</strong> <strong>space</strong> <strong>forms</strong> 39<br />
In C n the previous valuations satisfy B k,q (Ω) = Γ k,q (Ω) s<strong>in</strong>ce dβ k,q = dγ k,q . For ɛ ≠ 0 no<br />
form β k,q has the same differential as γ k,q .<br />
The differential of the <strong>forms</strong> θ 0 , θ 1 and θ 2 is given <strong>in</strong> [BF08] for ɛ = 0. From the structure<br />
equations <strong>in</strong> CK n (ɛ) (cf. [KN69]), <strong>in</strong> the same way as <strong>in</strong> [Par02], we compute the differential<br />
for any ɛ.<br />
Lemma 2.4.8. In S(CK n (ɛ)) it is satisfied<br />
dα = −θ s , dθ 0 = −ɛ(α ∧ θ 1 + βθ s ),<br />
dβ = θ 1 , dθ 1 = dθ 2 = dθ s = 0<br />
dγ = 2θ 0 − 2ɛ(α ∧ β + θ 2 ).<br />
In the follow<strong>in</strong>g proposition we give the relation between valuations {B k,q (Ω)} and {Γ k,q (Ω)}<br />
<strong>in</strong> CK n (ɛ).<br />
Proposition 2.4.9. In CK n (ɛ), for any pair of <strong>in</strong>tegers (k, q) such that max{0, k − n} < q <<br />
k/2 < n it is satisfied<br />
c n,k,q<br />
Γ k,q (Ω) = B k,q (Ω) − ɛ B k+2,q+1 (Ω)<br />
c n,k+2,q+1<br />
(q + 1)(2n − k)<br />
= B k,q (Ω) − ɛ<br />
2π(n − k + q) B k+2,q+1(Ω).<br />
Proof. Denote by I the ideal generated by α, dα and all the exact <strong>forms</strong> <strong>in</strong> N(Ω) (see Def<strong>in</strong>ition<br />
1.3.10). If two <strong>forms</strong> λ and ρ of degree 2n − 1 co<strong>in</strong>cide modulo I, then by Lemma 1.3.12<br />
∫ ∫<br />
λ = ρ.<br />
Thus, it is enough to prove<br />
N(Ω)<br />
c n,k,q<br />
N(Ω)<br />
γ k,q ≡ β k,q − ɛ β k+2,q+1 mod I. (2.7)<br />
c n,k+2,q+1<br />
Consider the form η = (θ s − β ∧ γ) ∧ θ n−k+q−1<br />
0 θ k−2q−1<br />
1 θ q 2 . As dη is exact we have dη ≡ 0<br />
mod I. On the other hand, from the differentials given <strong>in</strong> Lemma 2.4.8 we obta<strong>in</strong><br />
dη ≡ −γθ n−k+q−1<br />
0 θ k−2q<br />
1 θ q 2 + 2βθn−k+q 0 θ k−2q−1<br />
1 θ q 2 − 2ɛβθn−k+q−1 0 θ k−2q−1<br />
1 θ q+1<br />
2 mod I.<br />
Us<strong>in</strong>g the def<strong>in</strong>ition of γ k,q and β k,q we get the relation (2.7).<br />
Remark 2.4.10. For n = 2, 3, the previous relations are given <strong>in</strong> [Par02].<br />
By the relation <strong>in</strong> Proposition 2.4.9, we def<strong>in</strong>e the Hermitian <strong>in</strong>tr<strong>in</strong>sic volumes <strong>in</strong> CK n (ɛ).<br />
Def<strong>in</strong>ition 2.4.11. For max{0, k−n} ≤ q ≤ k 2<br />
< n, we def<strong>in</strong>e the Hermitian <strong>in</strong>tr<strong>in</strong>sic volumes<br />
µ k,q <strong>in</strong> CK n (ɛ)<br />
{<br />
Bk,q (Ω) si k ≠ 2q<br />
µ k,q (Ω) :=<br />
(2.8)<br />
Γ 2q,q (Ω) si k = 2q.<br />
Remark 2.4.12. In C n , Hermitian <strong>in</strong>tr<strong>in</strong>sic volumes form a basis of cont<strong>in</strong>uous valuations<br />
<strong>in</strong>variant under the isometry group of C n (cf. [BF08]).<br />
In the previous def<strong>in</strong>ition of µ k,q we take an arbitrary choice, despite of the relations <strong>in</strong><br />
Proposition 2.4.9. It would be <strong>in</strong>terest<strong>in</strong>g to know if there is a better choice, such that it<br />
satisfies some more natural geometric or algebraic properties.
40 Introduction to valuations<br />
2.4.3 Relation between Hermitian <strong>in</strong>tr<strong>in</strong>sic volumes and the valuations given<br />
by Park<br />
We give, by completeness, the valuations def<strong>in</strong>ed by Park <strong>in</strong> CK n (ɛ).<br />
Def<strong>in</strong>ition 2.4.13. Denote θ 00 = −α ∧ β + θ 2 , θ 01 = −β ∧ γ + θ s , θ 10 = −α ∧ γ + θ 1 , θ 11 = θ 0 .<br />
Let κ = σ ∧ {a, b, c} where σ ∈ {α, β} and {a, b, c} = 1<br />
a!b!c! θa 11 ∧ θb 00 ∧ θc 10 . Then<br />
∫<br />
Φ κ (Ω) =<br />
N(Ω)<br />
are smooth valuation <strong>in</strong>variant under the action of the isometry group of CK n (ɛ).<br />
The relation between these valuations and the Hermitian <strong>in</strong>tr<strong>in</strong>sic volumes is given straightforward<br />
from the def<strong>in</strong>ition of each valuation.<br />
Proposition 2.4.14. Let Ω ⊂ CK n (ɛ) be a regular doma<strong>in</strong>. Then,<br />
1<br />
B k,q (Ω) =<br />
Φ β{n−k+q,q,k−2q−1} (Ω),<br />
(k − 2q)ω 2n−k<br />
1<br />
Γ k,q (Ω) =<br />
Φ γ{n−k+q−1,q,k−2q} (Ω).<br />
2(n − k + q)ω 2n−k<br />
κ<br />
2.4.4 Other curvature <strong>in</strong>tegrals<br />
Let M be a Kähler manifold of <strong>complex</strong> dimension n and suppose that S ⊂ M is a real<br />
hypersurface. Then, we can canonically def<strong>in</strong>e a distribution of <strong>complex</strong> dimension n − 1 <strong>in</strong><br />
the tangent fiber bundle of S <strong>in</strong> the follow<strong>in</strong>g way.<br />
Let N x be the normal fiber bundle of S at x. Let J be the <strong>complex</strong> structure <strong>in</strong> M. The<br />
vector JN x is a tangent vector to S at x. Consider the orthogonal vectors to JN x <strong>in</strong>side the<br />
tangent <strong>space</strong> of S at x. These form a <strong>complex</strong> sub<strong>space</strong> of dimension n − 1. Denote by D<br />
the distribution def<strong>in</strong>ed by these sub<strong>space</strong>s. Then, we def<strong>in</strong>e the mean curvature <strong>in</strong>tegrals<br />
restricted to the distribution D as<br />
Def<strong>in</strong>ition 2.4.15. Let S be a hypersurface of a Kähler manifold M of <strong>complex</strong> dimension n.<br />
If x ∈ S, we denote the second fundamental form of S at x by II x and the second fundamental<br />
form restricted to D by II x | D . The r-th mean curvature <strong>in</strong>tegral of S restricted at D, 1 ≤ r ≤<br />
2n − 2, is def<strong>in</strong>ed as<br />
( ) 2n − 2 −1 ∫<br />
Mr D (S) =<br />
σ r (II x | D )dx<br />
r<br />
where σ r (II x | D ) denotes the r-th symmetric elementary function of II x | D .<br />
Along this work, we use the idea of restrict<strong>in</strong>g mean curvature <strong>in</strong>tegrals to the distribution<br />
D. The first mean curvature <strong>in</strong>tegral restricted to D will play an important role, i.e. the<br />
<strong>in</strong>tegral of the trace of the second fundamental form restricted to the distribution. Also the<br />
<strong>in</strong>tegral over the normal curvature JN will have an important role. If Ω is a regular doma<strong>in</strong>,<br />
we have the follow<strong>in</strong>g relations<br />
∫<br />
(2n − 1)M 1 (∂Ω) − (2n − 2)M1 D (∂Ω) =<br />
∂Ω<br />
S<br />
k n (JN)dp = 2ω 2 Γ 2n−2,n−1 (Ω). (2.9)
2.4 Valuations on <strong>complex</strong> <strong>space</strong> <strong>forms</strong> 41<br />
2.4.5 Relation between the Hermitian <strong>in</strong>tr<strong>in</strong>sic volumes and the second<br />
fundamental form<br />
In this section we give another expression for the Hermitian <strong>in</strong>tr<strong>in</strong>sic volumes <strong>in</strong> terms of the<br />
second fundamental form.<br />
In the proof of Theorem 4.3.1 we shall use some properties, <strong>in</strong>terest<strong>in</strong>g by their own, of<br />
the expression of the <strong>in</strong>variant (2n − 1)-<strong>forms</strong> expressed <strong>in</strong> terms of the second fundamental<br />
form of ∂Ω, and not only <strong>in</strong> terms of the connection <strong>forms</strong> of a mov<strong>in</strong>g frame. In order to give<br />
this expression <strong>in</strong> terms of the second fundamental form (with respect to a fixed basis) it is<br />
necessary to consider the pull-back of the follow<strong>in</strong>g canonical map<br />
ϕ : ∂Ω −→ N(Ω)<br />
x ↦→ (x, N x ) . (2.10)<br />
Let us study some properties of ϕ ∗ (β k,q ) and ϕ ∗ (γ k,q ).<br />
Let x ∈ ∂Ω ⊂ CK n (ɛ) and let {e 1 = ϕ(x), e 1 = Je 1 , . . . , e n , e n = Je n } be a J-mov<strong>in</strong>g<br />
frame def<strong>in</strong>ed <strong>in</strong> a neighborhood of x. Consider the 1-<strong>forms</strong> {α i , β i , α 1j , β 1j }, and the 2-<strong>forms</strong><br />
{θ 0 , θ 1 , θ 2 , θ s } given at (2.6).<br />
Notation 2.4.16. In order to simplify the notation <strong>in</strong> the follow<strong>in</strong>g expressions we denote β i<br />
by α i<br />
and β 1i by α 1i<br />
and we def<strong>in</strong>e I := {1, 2, 2, . . . , n, n}.<br />
Now, us<strong>in</strong>g the relation between the connection <strong>forms</strong> α 1i , i ∈ I, and the second fundamental<br />
form<br />
α 1i = ∑ h ij α i , (2.11)<br />
j∈I<br />
we obta<strong>in</strong><br />
Lemma 2.4.17. In the previous notation,<br />
ϕ ∗ (β) = α 1 ,<br />
ϕ ∗ (γ) = ∑ h 1j α j ,<br />
j∈I<br />
n∑ ∑<br />
ϕ ∗ (θ 0 ) = h ij h il<br />
α j ∧ α l ,<br />
ϕ ∗ (θ 1 ) =<br />
ϕ ∗ (θ 2 ) =<br />
i=2 j,l∈I<br />
⎛<br />
n∑<br />
⎝ ∑ h ij<br />
α i ∧ α j − ∑<br />
j∈I<br />
l∈I<br />
i=2<br />
n∑<br />
α i ∧ α i<br />
.<br />
i=2<br />
h il α i<br />
∧ α l<br />
⎞<br />
⎠ ,<br />
On the other hand, each form ϕ ∗ (β k,q ) is a form of maximum degree, and, thus, a multiple<br />
of the volume element dx = α 1 ∧ α 2 ∧ α 2 ∧ · · · ∧ α n of ∂Ω. Thus, ϕ ∗ (β k,q ) is determ<strong>in</strong>ed by this<br />
multiple, which can be <strong>in</strong>terpreted as a polynomial with variables the entries of the second<br />
fundamental form h ij (with respect to the fixed J-mov<strong>in</strong>g frame).<br />
In [Par02], it is computed explicitly the pull-back of the <strong>forms</strong> β k,q and γ k,q for dimensions<br />
n = 2, 3. In the follow<strong>in</strong>g lemma we give some general properties for the pull-back of these<br />
<strong>forms</strong> for any dimension n.<br />
Lemma 2.4.18. Let Ω ⊂ CK n (ɛ) be a regular doma<strong>in</strong> and let ϕ : ∂Ω → N(Ω) be the canonical<br />
map def<strong>in</strong>ed at (2.10). Let us fix a po<strong>in</strong>t x ∈ ∂Ω and a J-mov<strong>in</strong>g frame {e 1 = ϕ(x), e 1 =<br />
Je 1 , . . . , e n , e n = Je n } at x. If<br />
ϕ ∗ (β k,q ) = Q k,q dx,<br />
ϕ ∗ (γ k,q ) = P k,q dx
42 Introduction to valuations<br />
where dx is the volume element of ∂Ω, then<br />
1. Q k,q is a polynomial of degree 2n − k − 1 with variables the entries of the second fundamental<br />
form h ij = II(e i , e j ), i, j ∈ {2, 2, . . . , n};<br />
2. P k,q is a polynomial of degree 2n − k − 1 with variables the entries of the second fundamental<br />
form h ij = II(e i , e j ), i, j ∈ {1, 2, 2, . . . , n};<br />
3. each monomial of P k,q conta<strong>in</strong><strong>in</strong>g only entries of the form h ii also conta<strong>in</strong>s h 11 and<br />
exactly n + q − k − 1 factors of the form h jj h jj<br />
, i ∈ {1, 2, 2, . . . , n}, j ∈ {2, . . . , n};<br />
4. the polynomials P k,q and Q k,q can be written as a sum of m<strong>in</strong>ors of the second fundamental<br />
form with rank r = 2n − k − 1;<br />
5. among the m<strong>in</strong>ors described at 4. appear all m<strong>in</strong>ors centered at the diagonal with degree r<br />
conta<strong>in</strong><strong>in</strong>g h 11 for P k,q , and not conta<strong>in</strong><strong>in</strong>g h 11 for Q k,q . It also appears all non-centered<br />
m<strong>in</strong>ors such that the <strong>in</strong>dices of the rows and the <strong>in</strong>dices of the columns determ<strong>in</strong><strong>in</strong>g a<br />
m<strong>in</strong>or satisfy<br />
(a) conta<strong>in</strong> n − k + q <strong>in</strong>dices, <strong>in</strong> the case of Q k,q , and n − k + q − 1, <strong>in</strong> the case of P k,q ,<br />
such that, if the <strong>in</strong>dex j appears as an <strong>in</strong>dex <strong>in</strong> the rows (resp. columns), then the<br />
<strong>in</strong>dex j also appears as an <strong>in</strong>dex <strong>in</strong> the rows (resp. columns) of the m<strong>in</strong>or. We say<br />
that the <strong>in</strong>dex j is paired <strong>in</strong> the rows (or <strong>in</strong> the columns);<br />
(b) conta<strong>in</strong> k − 2q − 1 <strong>in</strong>dices non-paired neither <strong>in</strong> the rows nor <strong>in</strong> the columns for<br />
Q k,q , and k − 2q for P k,q ;<br />
(c) if the <strong>in</strong>dex j is <strong>in</strong> the non-paired <strong>in</strong>dices of the rows, then the <strong>in</strong>dex j is not <strong>in</strong> the<br />
<strong>in</strong>dex of the columns.<br />
Proof. From Lemma 2.4.17 we have<br />
and<br />
⎛<br />
⎞<br />
n∑ ∑<br />
ϕ ∗ (β k,q ) = c n,k,q α 1 ∧ ⎝ h ij h il<br />
α j ∧ α l<br />
⎠<br />
ϕ ∗ (γ k,q ) = c n,k,q<br />
2<br />
i=2 j,l∈I<br />
n+q−k<br />
⎛ ⎛<br />
⎞⎞<br />
n∑<br />
∧ ⎝ ⎝ ∑ h ij<br />
α i ∧ α j − ∑ h il α i<br />
∧ α l<br />
⎠⎠<br />
i=2 j∈I<br />
l∈I<br />
∧ (2.12)<br />
k−2q−1<br />
⎛ ⎞ ⎛<br />
⎞<br />
⎝ ∑ n+q−k−1<br />
n∑ ∑<br />
h 1j α j<br />
⎠ ∧ ⎝ h ij h il<br />
α j ∧ α l<br />
⎠ ∧<br />
j∈I<br />
i=2 j,l∈I<br />
⎛ ⎛<br />
⎞⎞<br />
n∑<br />
∧ ⎝ ⎝ ∑ h ij<br />
α i ∧ α j − ∑ h il α i<br />
∧ α l<br />
⎠⎠<br />
i=2 j∈I<br />
l∈I<br />
k−2q<br />
( n∑<br />
) q<br />
∧ α i ∧ α i<br />
,<br />
i=2<br />
( n∑<br />
) q<br />
∧ α i ∧ α i<br />
.<br />
i=2<br />
Thus, as ϕ ∗ (β k,q ) and ϕ ∗ (γ k,q ) are differential <strong>forms</strong> def<strong>in</strong>ed on ∂Ω of maximum degree, we<br />
have that ϕ ∗ (β k,q ) = Q k,q dx and ϕ ∗ (γ k,q ) = P k,q dx satisfy that Q k,p and P k,q are polynomials<br />
of degree 2n − k − 1. Moreover, polynomials P k,q cannot conta<strong>in</strong> h 11 s<strong>in</strong>ce this variable is<br />
multiplied by α 1 (see formula (2.11)), but this differential form is a common factor <strong>in</strong> the<br />
expression ϕ ∗ (β k,q ).
2.4 Valuations on <strong>complex</strong> <strong>space</strong> <strong>forms</strong> 43<br />
In order to prove 3. we observe that terms <strong>in</strong> the expression ϕ ∗ (γ k,q ) conta<strong>in</strong><strong>in</strong>g only entries<br />
of the type h ii are<br />
( n∑<br />
) n+q−k−1 ( n∑<br />
) k−2q ( n∑<br />
) q<br />
c n,k,q<br />
2 h 11 α 1 ∧ h ii h ii<br />
α i ∧ α i<br />
∧ h ii<br />
α i ∧ α i<br />
− h ii α i<br />
∧ α i ∧ α i ∧ α i<br />
.<br />
i=2<br />
i=2<br />
i=2<br />
So, the variable h 11 always appears and there are exactly n + q − k − 1 factors of the form<br />
h jj h jj<br />
, which come from ( ∑ n<br />
i=2 h iih ii<br />
α i ∧ α i<br />
) n+q−k−1 .<br />
A m<strong>in</strong>or of rank r of a matrix is def<strong>in</strong>ed choos<strong>in</strong>g r rows and r columns of the matrix and<br />
then tak<strong>in</strong>g the determ<strong>in</strong>ant of the square submatrix. To prove 4. and 5. we study <strong>in</strong> more<br />
detail the expression (2.12). (An analogous study can be done <strong>in</strong> the case ϕ ∗ (γ k,q ).) First, note<br />
that factors α 1 and ϕ ∗ (θ 2 ) do not give any term of the second fundamental form and, thus,<br />
they do not <strong>in</strong>fluence <strong>in</strong> the m<strong>in</strong>or (but they do <strong>in</strong>fluence <strong>in</strong> which m<strong>in</strong>ors can be constructed).<br />
Thus, we have to prove that the expression<br />
ϕ ∗ (θ n+q−k<br />
0 ) ∧ ϕ ∗ (θ k−2q−1<br />
1 ), (2.13)<br />
is a form of degree 2n−2q −2 where each term α i1 ∧α i2 ∧· · ·∧α i2n−2q−2 goes with a summation<br />
of m<strong>in</strong>ors.<br />
Develop<strong>in</strong>g (2.13) we have that it is equivalent to<br />
⎛<br />
⎞<br />
n∑<br />
n∑<br />
ϕ ∗ ⎝( α 1i ∧ α 1i<br />
) n+q−k ∧ ( (α j ∧ α 1j<br />
− α j ∧ α 1j )) k−2q−1 ⎠ .<br />
i=2<br />
j=2<br />
To develop this expression, first, we chose a := n + q − k values {i 1 , . . . , i a } for the <strong>in</strong>dex i<br />
of the first summation and b := k − 2q − 1 values {j 1 , . . . , j b } (with i k , j l ∈ {2, . . . , n}) for<br />
the <strong>in</strong>dex j of the second summation. Note that some <strong>in</strong>dexes can be repeated. Thus, we get<br />
( n−1<br />
n+q−k<br />
) (<br />
·<br />
n−1<br />
k−2q−1)<br />
summands<br />
ϕ ∗ (α 1i1 ∧ α 1i1<br />
∧ · · · ∧ α 1ia ∧ α 1ia<br />
∧ (α j1 ∧ α 1j1<br />
− α j1<br />
∧ α 1j1 ) ∧ · · · ∧ (α jb ∧ α 1jb<br />
− α jb<br />
∧ α 1jb )),<br />
which can be decomposed, for example, <strong>in</strong> the form<br />
ϕ ∗ (α 1i1 ∧ α 1i1<br />
∧ · · · ∧ α 1ia ∧ α 1ia<br />
∧ α j1 ∧ α 1j1<br />
∧ · · · ∧ α jb ∧ α 1jb<br />
)<br />
= α j1 ∧ · · · ∧ α jb ϕ ∗ (α 1i1 ∧ α 1i1<br />
∧ · · · ∧ α 1ia ∧ α 1ia<br />
∧ α 1j1<br />
∧ · · · ∧ α 1jb<br />
).<br />
From (2.11) the form <strong>in</strong> which we take pull-back can be expressed as the summation of the<br />
m<strong>in</strong>ors with rows given by the <strong>in</strong>dices I := {i 1 , i 1 , . . . , i a , i a , j 1 , . . . , j b }, and by columns each<br />
of the possible permutations of the elements without repetition among the <strong>in</strong>dexes <strong>in</strong> J :=<br />
I \ {1, j 1 , . . . , j b }. Note that <strong>in</strong>dex α 1 cannot be taken s<strong>in</strong>ce we are consider<strong>in</strong>g the form <strong>in</strong><br />
(2.12), which is multiplied by α 1 . (In the case of ϕ ∗ (γ k,q ) we do not have this restriction, and,<br />
so, can also appear m<strong>in</strong>ors with the term h 11 .)<br />
If we chose for J the same <strong>in</strong>dexes as <strong>in</strong> I, then we get all the m<strong>in</strong>ors centered at the<br />
diagonal. Condition 5.(c) is obta<strong>in</strong>ed directly from the fact that the <strong>in</strong>dices which determ<strong>in</strong>es<br />
the columns have to be <strong>in</strong> J , and if j k is an <strong>in</strong>dex <strong>in</strong> the rows, the <strong>in</strong>dex j k is not <strong>in</strong> J .<br />
Conditions 5.(a) and 5.(b) are obta<strong>in</strong>ed when we recall that we are not study<strong>in</strong>g the differential<br />
form <strong>in</strong> (2.13) but <strong>in</strong> (2.12), i.e. we have to take the product with α 1 ∧ ( ∑ n<br />
i=2 α i ∧ α i<br />
) q .<br />
But, if this product conta<strong>in</strong>s the form α k , then it also conta<strong>in</strong>s the form α k<br />
(except for k = 1).<br />
Thus, to complete the form <strong>in</strong> (2.13) to a (2n − 1)-differential form we have to take the products<br />
α k ∧ α k<br />
, so that, <strong>in</strong> order to not obta<strong>in</strong> a vanish<strong>in</strong>g term, the quantity of paired <strong>in</strong>dices<br />
<strong>in</strong> the rows and <strong>in</strong> the columns must be the same, and, thus, it also co<strong>in</strong>cides the quantity of<br />
no-paired <strong>in</strong>dices.
44 Introduction to valuations<br />
Remarks 2.4.19. 1. Each of the m<strong>in</strong>or goes with a constant depend<strong>in</strong>g on the number of<br />
permutations that allows us to obta<strong>in</strong> it. Thus, not all the m<strong>in</strong>ors have the same constant,<br />
but, for <strong>in</strong>stance all the m<strong>in</strong>ors (fixed β k,q or γ k,q ) centered at the diagonal have the same<br />
one.<br />
2. The degree of the polynomial given by β k,q or γ k,q does not depend on q, just on k, but<br />
two polynomials com<strong>in</strong>g from β k,q and β k,q ′ (or γ k,q and γ k,q ′) are dist<strong>in</strong>guished by the<br />
number of paired <strong>in</strong>dexes.<br />
Example 2.4.20. We give the explicit relation between some Hermitian <strong>in</strong>tr<strong>in</strong>sic volumes and<br />
the second fundamental form.<br />
1. ϕ ∗ (γ 00 ) = c n,0,0<br />
2 ϕ∗ (γ ∧ θ0 n−1 ) = (n − 1)! c n,0,0<br />
det(II)dx = 1 det(II)dx.<br />
2<br />
2nω 2n<br />
2. ϕ ∗ (β 10 ) = c n,1,0 ϕ ∗ (β ∧ θ n−1<br />
0 ) = (n − 1)!c n,1,0 det(II| D )dx = 1<br />
ω 2n−1<br />
det(II| D )dx.<br />
3. ϕ ∗ (γ 2n−2,n−1 ) = c n,2n−2,n−1<br />
ϕ ∗ (γ∧θ n−1<br />
2<br />
4. ϕ ∗ (β 2n−2,n−2 ) = c n,2n−2,n−2 ϕ ∗ (β ∧ θ 1 ∧ θ n−2<br />
2 ) =<br />
2 ) = c n,2n−1,n−1(n − 1)!<br />
2<br />
k n (JN)dx = k n (JN) dx<br />
2ω 2<br />
.<br />
(n − 2)!<br />
(n − 2)!2ω 2<br />
tr(II| D )dx = tr(II| D ) dx<br />
2ω 2<br />
.<br />
5. ϕ ∗ (β 2n−1,n−1 ) = c n,2n−1,n−1 ϕ ∗ (β ∧ θ n−1<br />
2 ) = c n,2n−1,n−1 (n − 1)!α 1 ∧ α 1 ∧ · · · ∧ α n = dx 2 .
Chapter 3<br />
Average of the mean curvature<br />
<strong>in</strong>tegral<br />
For the real <strong>space</strong> <strong>forms</strong> (R n , S n and H n ), it is known that the reproductive property holds for<br />
mean curvature <strong>in</strong>tegrals. That is, given a regular doma<strong>in</strong> Ω, it is satisfied (cf. Example 2.2.3)<br />
∫<br />
M r<br />
(s)<br />
L s<br />
(∂Ω ∩ L s )dL s = cM r (∂Ω).<br />
On the other hand, by Section 2.3, this property may not hold <strong>in</strong> C n , when we <strong>in</strong>tegrate over<br />
the <strong>space</strong> of <strong>complex</strong> planes. Thus, it is natural to study, <strong>in</strong> C n , the value of<br />
∫<br />
L C s<br />
M (s)<br />
r (∂Ω ∩ L s )dL s .<br />
In the same way, we will study the value of this <strong>in</strong>tegral but <strong>in</strong> the other <strong>complex</strong> <strong>space</strong><br />
<strong>forms</strong>, CP n and CH n . Recall that we denote by CK n (ɛ) the <strong>space</strong> of constant holomorphic<br />
curvature 4ɛ.<br />
In this chapter we deduce the expression of the <strong>in</strong>tegral of M (s)<br />
1 (∂Ω ∩ L s) <strong>in</strong> terms of the<br />
mean curvature <strong>in</strong>tegral of the convex doma<strong>in</strong> M 1 (∂Ω) and the <strong>in</strong>tegral of the normal curvature<br />
<strong>in</strong> the direction JN, ∫ ∂Ω k n(JN) (see Theorem 3.3.2). We also f<strong>in</strong>d a partial result for the<br />
<strong>in</strong>tegral over any other mean curvature <strong>in</strong>tegral M r<br />
(s) (∂Ω∩L s ), 0 ≤ r ≤ 2s−1 (see Proposition<br />
3.2.2).<br />
3.1 Previous lemmas<br />
First of all, we state some lemmas that will be necessary <strong>in</strong> order to prove Theorem 3.3.2 and<br />
some other results.<br />
Lemma 3.1.1. Let V be a <strong>complex</strong> vector <strong>space</strong> of <strong>complex</strong> dimension 2 endowed with an <strong>in</strong>ner<br />
product 〈 , 〉 compatible with the <strong>complex</strong> structure J and let {e 1 , e 2 , e 3 , e 4 } be an orthonormal<br />
basis of V . Then 〈e a , Je b 〉 2 = 〈e c , Je d 〉 2 with {a, b, c, d} = {1, 2, 3, 4}.<br />
Proof. We express Je b and Je d <strong>in</strong> terms of the orthonormal, and we obta<strong>in</strong><br />
Je b = 〈Je b , e a 〉e a + 〈Je b , e c 〉e c + 〈Je b , e d 〉e d = Ae a + Be c + Ce d ,<br />
Je d = 〈Je d , e a 〉e a + 〈Je d , e b 〉e b + 〈Je d , e c 〉e c = De a + Ee b + F e c .<br />
45
46 Average of the mean curvature <strong>in</strong>tegral<br />
Now, us<strong>in</strong>g that 〈Je b , Je b 〉 = 〈Je d , Je d 〉 = 1, 〈Je b , Je d 〉 = 0 and 〈Je b , e d 〉 = −〈Je d , e b 〉, we<br />
get<br />
and<br />
A 2 + B 2 + C 2 = 1,<br />
D 2 + E 2 + F 2 = 1,<br />
AD + BF = 0,<br />
C = −E,<br />
A 2 + B 2 = D 2 + F 2 and A 2 D 2 = B 2 F 2 .<br />
F<strong>in</strong>ally, we substitute D 2 = A 2 + B 2 − F 2 <strong>in</strong> the second equation and we obta<strong>in</strong> A 2 = F 2 .<br />
Lemma 3.1.2. If u ∈ S 2n−3 , then<br />
∫<br />
S 2n−3 〈u, v〉dv = 0<br />
and<br />
∫<br />
S 2n−3 〈u, v〉 2 dv = ω 2n−2 .<br />
Proof. The first equality follows s<strong>in</strong>ce the <strong>in</strong>tegral is over an odd function. For the second one,<br />
we decompose v = cos θu + s<strong>in</strong> θw with w ∈ 〈u〉 ⊥ , then us<strong>in</strong>g polar coord<strong>in</strong>ates with respect<br />
to u, we have<br />
∫S 2n−3 〈u, v〉 2 dv = O 2n−4<br />
∫ π<br />
0<br />
cos 2 θ s<strong>in</strong> 2n−4 θdθ = O 2n−4<br />
O 2n−4+2+1<br />
O 2 O 2n−4<br />
= ω 2n−2 .<br />
where O n denotes the volume of the n-dimensional Euclidean sphere and ω n the volume of the<br />
the n-dimensional Euclidean ball.<br />
The follow<strong>in</strong>g lemma gives a generalized version of the Meus<strong>in</strong>er Theorem.<br />
Lemma 3.1.3. Let S ⊂ M be a hypersurface of class C 2 of a Riemannian manifold M, p ∈ S<br />
and L ⊂ T p M a vector sub<strong>space</strong>. We denote by II S the second fundamental form of S and<br />
by II L C the second fundamental form of C = S ∩ exp p L as a hypersurface of exp p L. We also<br />
denote u = T p S ∩ L. Then,<br />
σ i (II S | u ) = cos i θσ i (II| L C)<br />
where θ denotes the angle at p between a normal vector of S and a normal vector of C <strong>in</strong><br />
exp p L, and σ i (Q) denotes the i-th symmetric elementary function of the bil<strong>in</strong>eal form Q.<br />
Proof. If A ⊂ B ⊂ M are submanifolds, then we denote the second fundamental form of A as<br />
a submanifold of B by h B A : T pA × T p A → (T p A) ⊥ . If B = M, we just put h A <strong>in</strong>stead of h M A .<br />
Then, for all X, Y ∈ T p C<br />
h C (X, Y ) = h L C(X, Y ) + h L (X, Y ) = h L C(X, Y )<br />
s<strong>in</strong>ce the second fundamental form of L vanishes at p. Moreover,<br />
h C (X, Y ) = h S C(X, Y ) + h S (X, Y ).<br />
Let N be a normal vector to S. Note that h S C<br />
(X, Y ) is a multiple of a normal vector to C <strong>in</strong><br />
S, so 〈h S C (X, Y ), N〉 = 0 (for X, Y ∈ T pC).<br />
If X, Y ∈ T p C, then<br />
II S (X, Y ) := 〈h S (X, Y ), N〉 = 〈h C (X, Y ) − h S C(X, Y ), N〉<br />
= 〈h C (X, Y ), N〉 = 〈h L C(X, Y ), N〉 = 〈II L C(X, Y )n, N〉
3.1 Previous lemmas 47<br />
where n denotes a normal vector of C <strong>in</strong> L. So,<br />
II L C(X, Y ) = 1<br />
〈N, n〉 II S(X, Y ). (3.1)<br />
S<strong>in</strong>ce σ i (II L C ) is the sum of the m<strong>in</strong>ors of order i of IIL C , by replac<strong>in</strong>g by (3.1) each entry of the<br />
second fundamental form, we obta<strong>in</strong> the result.<br />
The follow<strong>in</strong>g lemma generalizes <strong>in</strong> C n a result given by Langev<strong>in</strong> and Shifr<strong>in</strong> [LS82] <strong>in</strong><br />
R n .<br />
Lemma 3.1.4. Let E be a <strong>complex</strong> vector <strong>space</strong> of <strong>complex</strong> dimension n and let II be a real<br />
bil<strong>in</strong>ear form def<strong>in</strong>ed on E. We denote by G C n,s the Grassmanian of s-dimensional <strong>complex</strong><br />
planes on E. Then,<br />
∫<br />
tr(II| V )dV = s vol(GC n,s)<br />
tr(II| E ).<br />
n<br />
G C n,s<br />
Proof. First, recall that<br />
is a fibration for each s ∈ {1, . . . , n − 1}.<br />
U(n − s) × U(s) −→ U(n) −→ G C n,s (3.2)<br />
We prove the case dim C V ≤ n 2<br />
by <strong>in</strong>duction on the <strong>complex</strong> dimension of V . The case<br />
dim C V > n 2<br />
can be proved us<strong>in</strong>g similar arguments.<br />
Suppose dim C V = 1, that is, s = 1. Then,<br />
∫<br />
G C n,1<br />
tr(II| V )dV =<br />
∫<br />
1<br />
tr(II|<br />
vol(U(n − 1))vol(U(1))<br />
V 1<br />
1<br />
)dU<br />
U(n)<br />
s<strong>in</strong>ce tr(II| V 1<br />
1<br />
) is constant along the fiber. We denote by V1 1 the <strong>complex</strong> vector sub<strong>space</strong><br />
generated by the first column of the matrix U ∈ U(n). In general, for U ∈ U(n), we will<br />
denote by Va b the <strong>complex</strong> vector sub<strong>space</strong> generated by the columns b to b + a − 1. The<br />
subscript a denotes the dimension of Va b , or equivalently, the number of columns we consider<br />
and the upperscript b denotes from which column we start to consider them. Then<br />
∫<br />
U(n)<br />
tr(II| V 1<br />
1<br />
)dU = 1 n<br />
= 1 n<br />
∫<br />
∫<br />
U(n)<br />
U(n)<br />
(tr(II| V 1<br />
1<br />
) + tr(II| V 2<br />
1<br />
) + · · · + tr(II| V n<br />
1<br />
))dU<br />
tr(II| E )dU = vol(U(n)) tr(II| E ).<br />
n<br />
Thus,<br />
∫<br />
G C n,1<br />
tr(II| V )dV =<br />
vol(U(n))<br />
nvol(U(n − 1))vol(U(1)) tr(II| E) = vol(GC n,1 ) tr(II| E ).<br />
n<br />
Suppose now that the result is true till dim C V = r −1. We shall prove it for dim C V = r ≤<br />
n<br />
2 . If R denotes the rema<strong>in</strong>der of n r<br />
, then R < r and we can apply the <strong>in</strong>duction hypothesis <strong>in</strong>
48 Average of the mean curvature <strong>in</strong>tegral<br />
R at equality (*). Thus, us<strong>in</strong>g similar arguments as before, we obta<strong>in</strong><br />
∫<br />
G C n,r<br />
tr(II| V ) =<br />
∫<br />
1<br />
tr(II|<br />
vol(U(n − r))vol(U(r))<br />
V 1 r<br />
)<br />
U(n)<br />
1<br />
(tr(II| V 1 r<br />
) + tr(II| V 2 r<br />
) + · · · + tr(II| V<br />
n−R−r+1<br />
=<br />
vol(U(n − r))vol(U(r))⌊ n r<br />
∫U(n) ⌋<br />
( ∫<br />
1<br />
=<br />
vol(U(n − r))vol(U(r))⌊ n r ⌋ U(n)<br />
(∗)<br />
=<br />
∫<br />
tr(II| E ) −<br />
U(n)<br />
tr(II| V<br />
n−R+1)<br />
R<br />
vol(U(n))tr(II| E ) − vol(U(n − R))vol(U(R)) ∫ G<br />
tr(II| C VR )<br />
n,R<br />
=<br />
vol(U(n − r))vol(U(r))⌊ n r<br />
(<br />
⌋ )<br />
vol(U(n)) − vol(U(n − R))vol(U(R))vol(G C n,R ) R n<br />
tr(II| E )<br />
= vol(G C r<br />
n,r)<br />
n − R<br />
= vol(G C n,r) r n tr(II| E)<br />
and the result follows when 2s ≤ n.<br />
vol(U(n − r))vol(U(r))⌊ n r<br />
(<br />
⌋<br />
1 − R )<br />
tr(II| E )<br />
n<br />
)<br />
r<br />
))<br />
3.2 <strong>Integral</strong> of the r-th mean curvature <strong>in</strong>tegral over the <strong>space</strong><br />
of <strong>complex</strong> s-planes<br />
Along this chapter we follow some conventions which we state <strong>in</strong> the follow<strong>in</strong>g paragraphs.<br />
We denote by S ⊂ CK n (ɛ) a hypersurface of class C 2 , compact and oriented (possibly with<br />
boundary). Given a <strong>complex</strong> s-plane L s <strong>in</strong>tersect<strong>in</strong>g S, it is said that L s is <strong>in</strong> generic position<br />
if S ∩ L s is a submanifold of dimension 2s − 1 <strong>in</strong> CK n (ɛ). For hypersurfaces of class C 2 , the<br />
subset of generic planes (<strong>in</strong>tersect<strong>in</strong>g S) has full measure. Thus, we suppose that each <strong>complex</strong><br />
s-plane is <strong>in</strong> generic position. Note that S ∩ L s (if L s is a <strong>complex</strong> s-plane <strong>in</strong> generic position)<br />
is a hypersurface <strong>in</strong> L s<br />
∼ = CK s (ɛ).<br />
Suppose that N denotes a unit normal vector field on S. In this chapter we take, <strong>in</strong> S ∩ L s<br />
as a submanifold <strong>in</strong> S, the normal vector field Ñ such that the angle between N and Ñ is<br />
acute. Along the proofs <strong>in</strong> this chapter, we denote e s := ±JÑ.<br />
Note that if p ∈ S ∩ L s and Ñp is the chosen normal vector field <strong>in</strong> S ∩ L s <strong>in</strong>side L s then<br />
JÑ ∈ T p(S ∩ L s ). Indeed, L s<br />
∼ = CK s (ɛ), thus, the same structure for hypersurfaces hols <strong>in</strong>side<br />
L s .<br />
We denote by E ⊂ T p CK n (ɛ), p ∈ S ∩ L s , the orthogonal sub<strong>space</strong> to the <strong>space</strong> generated<br />
by {N, JN, Ñ, JÑ}. Note that exp p(E) ∼ = CK n−2 (ɛ) and it is univocally determ<strong>in</strong>ed for each<br />
L s .<br />
The fact stated <strong>in</strong> the follow<strong>in</strong>g remark is used implicitly along this chapter, specially to<br />
def<strong>in</strong>e the mov<strong>in</strong>g frames g and g ′ <strong>in</strong> the proof of the next proposition.<br />
Remark 3.2.1. Let V n be an n-dimensional Hermitian <strong>space</strong> with <strong>complex</strong> structure J, H ⊂ V n<br />
a real hyperplane and W s ⊂ V n a <strong>complex</strong> sub<strong>space</strong> of dimension s.<br />
Consider the sub<strong>space</strong> H ∩ W s and denote by N ′ an orthonormal vector to H ∩ W s <strong>in</strong> W s ,<br />
and D ′ = 〈N ′ , JN ′ 〉 ⊥ ∩ W s .<br />
Consider also the sub<strong>space</strong>s H ∩ Ws<br />
⊥<br />
<strong>in</strong> Ws ⊥ , and D ′′ = 〈N ′′ , JN ′′ 〉 ⊥ ∩ W s ′′ .<br />
and denote by N ′′ an orthonormal vector to H ∩ W ⊥ s
3.2 <strong>Integral</strong> of the r-th mean curvature <strong>in</strong>tegral 49<br />
Denote by N an orthonormal vector to H <strong>in</strong> V , and D = 〈N, JN〉 ⊥ . Then,<br />
V n = D⊥〈JN〉 R ⊥〈N〉 R<br />
= D ′ ⊥〈JN ′ 〉 R ⊥〈N ′ 〉 R ⊥D ′′ ⊥〈JN ′′ 〉 R ⊥〈N ′′ 〉 R .<br />
Indeed, from section 2.4.4 given a real hyperplane W <strong>in</strong> a Hermitian <strong>space</strong> V there exists<br />
a canonical decomposition of V as<br />
V = D⊥〈JN〉 R ⊥〈N〉 R ,<br />
where N is an orthogonal vector to W <strong>in</strong> V . Apply<strong>in</strong>g this fact to V = W s and to V = W ⊥ s<br />
we get the result.<br />
The follow<strong>in</strong>g proposition shall be essential to prove Theorem 3.3.2 and other results,<br />
s<strong>in</strong>ce it gives a first expression of the <strong>in</strong>tegral over the <strong>space</strong> of <strong>complex</strong> s-planes of the mean<br />
curvature <strong>in</strong>tegral <strong>in</strong> terms of an <strong>in</strong>tegral on the boundary of the doma<strong>in</strong>.<br />
Proposition 3.2.2. Let S ⊂ CK n (ɛ) be a compact (possibly with boundary) hypersurface of<br />
class C 2 oriented by a normal vector N, and let r, s ∈ N such that 1 ≤ s ≤ n and 0 ≤ r ≤ 2s−1.<br />
Then<br />
∫<br />
( ) 2s − 1 −1 ∫ ∫ ( ∫ )<br />
M r (s)<br />
|〈JN, e s 〉|<br />
(S∩L s )dL s =<br />
L C r<br />
s<br />
S( 2s−r<br />
RP 2n−2 G C (1 − 〈JN, e<br />
n−2,s−1<br />
s 〉 2 ) s−1 σ r(p; e s ⊕ V )dV<br />
)de s dp,<br />
where e s ∈ T p S unit vector, V denotes a <strong>complex</strong> (s−1)-plane by p conta<strong>in</strong>ed <strong>in</strong> {N, JN, e s , Je s } ⊥ ,<br />
σ r (p; e s ⊕ V ) denotes the r-th symmetric elementary function of the second fundamental form<br />
of S restricted to the real sub<strong>space</strong> e s ⊕V and the <strong>in</strong>tegration over RP 2n−2 denotes the projective<br />
<strong>space</strong> of the unit tangent <strong>space</strong> of the hypersurface.<br />
Remark 3.2.3. Us<strong>in</strong>g the previous remarks, it follows that the product |〈JN, e s 〉| <strong>in</strong> the last<br />
proposition gives the cos<strong>in</strong>e of the acute angle between the normal vector to the hypersurface<br />
S <strong>in</strong> CK n (ɛ) and a normal vector to S ∩ L s <strong>in</strong> L s , that is,<br />
|〈JN, e s 〉| = |〈N, Ñ〉|.<br />
Proof. Let L s be a <strong>complex</strong> s-plane such that S ∩ L s ≠ ∅ and let p ∈ S ∩ L s . We denote<br />
by ˜σ r the r-th symmetric elementary function of the second fundamental form of S ∩ L s as a<br />
hypersurface of L s . Then, by def<strong>in</strong>ition<br />
∫<br />
( ) 2s − 1 −1 ∫ ∫<br />
M r (s) (S ∩ L s )dL s =<br />
˜σ r (s)dx dL s .<br />
r S∩L s≠∅ S∩L s<br />
L C s<br />
We shall prove the result us<strong>in</strong>g mov<strong>in</strong>g frames adapted to S ∩ L s , L s or S.<br />
Let g = {e 1 , e 1 = Je 1 , e 2 , e 2 = Je 2 , . . . , e s , w s , e s+1 , e s+1 = Je s+1 . . . , e n , N} be a mov<strong>in</strong>g<br />
frame adapted to S ∩ L s and S (cf. Remark 3.2.1). That is, {e 1 , e 1 , . . . , e s } is an orthonormal<br />
basis of T p (S ∩L s ), {e s+1 , e s+1 , . . . , e n } is an orthonormal basis of T p S ∩(T p L s ) ⊥ , N is a normal<br />
vector field to T S and w s completes to an orthonormal basis of T p CK n (ɛ). We denote by<br />
{ω 1 , ω 1 , . . . , ω s−1 , ω s−1 , ω s , ω˜s , ω s+1 , ω s+1 , . . . , ω n , ωñ}<br />
the dual basis of the vectors <strong>in</strong> g and by {ω ij }, i, j ∈ {1, 1, . . . , s, ˜s, s + 1, s + 1, . . . , n, ñ}, the<br />
connection <strong>forms</strong> (cf. (1.15)).<br />
Let g ′ = {e ′ 1 = e 1, e ′ 1 = Je 1, e ′ 2 = e 2, e ′ 2 = Je 2, ..., e ′ s = e s , e ′ s = Je s, e ′ s+1 = e s+1, e ′ s+1 =<br />
Je s+1 , ..., e ′ n = e n , e ′ n = Je n} be a mov<strong>in</strong>g frame adapted to S ∩ L s and L s . That is,
50 Average of the mean curvature <strong>in</strong>tegral<br />
{e ′ 1 , e′ 1 , ..., e′ s} is an orthonormal basis of T p (L s ∩ S) and {e ′ 1 , e′ 1 , . . . , e′ s, e s ′ } is an orthonormal<br />
basis of T p L s . Denote by<br />
{ω 1, ′ ω ′ 1 , . . . , ω′ n, ω n}<br />
′<br />
the dual basis of vectors <strong>in</strong> g ′ and by {ω ij ′ } the connection <strong>forms</strong>, i, j ∈ {1, 1, . . . , n, n}.<br />
As the base g and g ′ are constituted by orthonormal vectors, we can easily give the relation<br />
among the elements <strong>in</strong> the frame g ′ and the ones <strong>in</strong> g just express<strong>in</strong>g the vectors <strong>in</strong> g ′ <strong>in</strong><br />
coord<strong>in</strong>ates with respect to the vectors <strong>in</strong> g:<br />
e ′ s = Je s = 〈Je s , w s 〉w s + 〈Je s , N〉N,<br />
e ′ n = Je n = 〈Je n , w s 〉w s + 〈Je n , N〉N<br />
and e ′ j = e j when j ∈ {1, 1, . . . , s − 1, s − 1, s, s + 1, s + 1, . . . , n − 1, n − 1, n}.<br />
Then { ω<br />
′<br />
j = ω j , if j ≠ s, n,<br />
and<br />
⎧<br />
⎨<br />
⎩<br />
ω ′ n<br />
= 〈Je n , w s 〉ω˜s + 〈Je n , N〉ωñ<br />
ω is ′ = 〈Je s , w s 〉ω i˜s + 〈Je s , N〉ω iñ , if i ≠ s, n<br />
ω <strong>in</strong> ′ = 〈Je n , w s 〉ω i˜s + 〈Je n , N〉ω iñ , if i ≠ s, n<br />
ω ij ′ = ω ij , if i, j ≠ s, n.<br />
(3.3)<br />
(3.4)<br />
From now on, <strong>in</strong> order to simplify the notation, we omit the absolute value <strong>in</strong> densities.<br />
The expression of dx (the density of S ∩ L s ), dL s and dL s[p] <strong>in</strong> terms of ω ′ is<br />
dx = ω ′ 1 ∧ ω ′ 1 ∧ · · · ∧ ω′ s,<br />
dL s = ω ′ s+1 ∧ ω ′ s+1 ∧ · · · ∧ ω′ n ∧ ω ′ n ∧<br />
dL s[p] =<br />
∧<br />
ω ij<br />
′<br />
i=1,2,...,s<br />
j=s+1,s+1,...,n,n<br />
∧<br />
ω ′ ij,<br />
i=1,2,...,s<br />
j=s+1,s+1,...,n,n<br />
and the expression of dp (the density of S) <strong>in</strong> terms of ω is dp = ω 1 ∧ ω 1 ∧ · · · ∧ ω n .<br />
On the other hand, by Lemma 3.1.1 it is satisfied<br />
|〈Je n , w s 〉| = |〈Je s , N〉|. (3.5)<br />
Indeed, vectors {e s , w s , e n , N} are an orthonormal basis of a 2-dimensional <strong>complex</strong> plane, the<br />
orthogonal complement of the <strong>space</strong> generated by {e 1 , Je 1 , . . . , e s−1 , Je s−1 , e s+1 , Je s+1 , . . . , e n−1 , Je n−1 }.<br />
By relations (3.3) and (3.5) we get<br />
s<strong>in</strong>ce ωñ vanishes on T S.<br />
Then, by Lemma 3.1.3,<br />
∫<br />
M (s)<br />
S∩L s≠∅<br />
dx ∧ dL s = |〈Je n , w s 〉|dL s[p] ∧ dp = |〈JN, e s 〉|dL s[p] ∧ dp (3.6)<br />
( ) 2s − 1 −1 ∫<br />
r (S ∩ L s )dL s =<br />
r<br />
( ) 2s − 1 −1 ∫<br />
=<br />
r<br />
S<br />
S<br />
∫<br />
|〈JN, e s 〉|˜σ r (p)dL s[p] dp<br />
L s[p]<br />
∫<br />
|〈JN, e s 〉|<br />
L s[p]<br />
|〈N, Je s 〉| r σ r(p)dL s[p] dp.<br />
Note that <strong>in</strong> the last <strong>in</strong>tegrand we consider the absolut value <strong>in</strong> the denom<strong>in</strong>ator to be<br />
sure that we consider the acute angle between the two <strong>in</strong>tersect<strong>in</strong>g sub<strong>space</strong>. This is not a
3.2 <strong>Integral</strong> of the r-th mean curvature <strong>in</strong>tegral 51<br />
priori true s<strong>in</strong>ce e s is any vector and, thus, not always the vector Je s (a normal vector to<br />
S ∩ L s ⊂ L s ) has the desired orientation.<br />
Now, we shall express dL s[p] <strong>in</strong> terms of dV ∧ de s where dV denotes the volume element of<br />
the Grassmannian G C n−2,s−1 and de s the volume element of S 2n−2 . Fixed p, def<strong>in</strong>e the manifold<br />
M = {(e s , V ) | e s ∈ T p S unit , V ∈ L s conta<strong>in</strong><strong>in</strong>g p and orthogonal to {N, JN, e s , Je s }},<br />
which locally co<strong>in</strong>cides with S 2n−2 × G C n,s−1 . Consider the fiber bundle<br />
φ : M −→ L C s[p]<br />
(e s , V ) ↦→ exp p {e s , Je s , V }<br />
with fiber G C n,s. This is a double cover<strong>in</strong>g of L C s[p]<br />
s<strong>in</strong>ce vectors v and −v give the same <strong>complex</strong><br />
s-plane.<br />
The pull-back of dL s[p] by the last mapp<strong>in</strong>g give the desired relation among the densities.<br />
The expression of de s <strong>in</strong> terms of ω and the expression of dV <strong>in</strong> terms of ω ′ are<br />
∧<br />
de s =<br />
ω sj ,<br />
By (3.4) and (3.7) we have<br />
∧<br />
dL s[p] =<br />
dV =<br />
ω ij<br />
′<br />
i=1,2,...,s<br />
j=s+1,s+1,...,n,n<br />
= dV ∧<br />
∧<br />
i=1,2,...,s−1<br />
j=1,1,...,s−1,s−1,˜s,s+1,s+1,...,n<br />
∧<br />
i=1,2,...,s−1<br />
j=s+1,s+1,...,n−1,n−1<br />
ω <strong>in</strong> ∧<br />
∧<br />
i=1,2,...,s<br />
ω ′ ij. (3.7)<br />
ω ′ <strong>in</strong> ∧<br />
Next, we relate ∧ ω <strong>in</strong> ′ ∧ ω<br />
′<br />
<strong>in</strong> with ∧ ω sj . From (3.4) follows<br />
∧<br />
∧<br />
ω <strong>in</strong> ′ = |〈Je n , w s 〉| s<br />
i=1,2,...,s<br />
i=1,2,...,s<br />
∧<br />
j=s+1,s+1,...,n−1,n<br />
and also us<strong>in</strong>g that ω <strong>in</strong> ′ = ω′ <strong>in</strong> s<strong>in</strong>ce ω′ <strong>in</strong> = 〈de′ i , e′ n〉 = 〈dJe ′ i , Je′ n〉 = ω ′ we obta<strong>in</strong><br />
i,n<br />
∧<br />
∧<br />
ω <strong>in</strong> ′ = ω ′ <strong>in</strong> =<br />
∧<br />
〈Je n , w s 〉ω i˜s<br />
In order to study<br />
i=1,2,...,s−1<br />
i=1,2,...,s−1<br />
= |〈Je n , w s 〉| s−1 ∧<br />
∧<br />
i=1,1,...,s−1,s−1<br />
i=1,2,...,s−1<br />
i=1,2,...,s−1<br />
ω i˜s ,<br />
we use e s = Je s = 〈Je s , w s 〉w s + 〈Je s , N〉N and we obta<strong>in</strong><br />
∧<br />
∧<br />
ω is =<br />
ω is<br />
=<br />
i=1,1,...,s−1,s−1<br />
i=1,1,...,s−1,s−1<br />
= 〈Je s , w s 〉 2(s−1) ∧<br />
ω i˜s<br />
.<br />
ω i˜s<br />
∧<br />
i=1,1,...,s−1,s−1<br />
i=1,1,...,s−1,s−1<br />
ω i˜s .<br />
ω is<br />
ω sj .
52 Average of the mean curvature <strong>in</strong>tegral<br />
Thus,<br />
and<br />
∧<br />
i=1,1,...,s−1,s−1<br />
ω i˜s = 〈Je s , w s 〉 −2(s−1)<br />
∧<br />
i=1,1,...,s−1,s−1<br />
dL s[p] = |〈Je n, w s 〉| 2s−1<br />
〈Je s , w s 〉 2(s−1) dV ∧ de s. (3.8)<br />
Us<strong>in</strong>g |〈Je n , w s 〉| = |〈JN, e s 〉| and 〈Je s , w s 〉 2 = 1 − 〈JN, e s 〉 2 we get the result.<br />
3.3 Mean curvature <strong>in</strong>tegral<br />
First we give an expression for the <strong>in</strong>tegral over the <strong>space</strong> of <strong>complex</strong> r-planes of the <strong>in</strong>tegral<br />
of the normal curvature <strong>in</strong> the direction JN (see (2.9)). By Example 2.4.20 we have that this<br />
<strong>in</strong>tegral is a smooth valuation <strong>in</strong> CK n (ɛ). Moreover, it is not a multiple of the mean curvature<br />
<strong>in</strong>tegral.<br />
Theorem 3.3.1. Let S ⊂ CK n (ɛ) be compact oriented (possibly with boundary) hypersurface<br />
of class C 2 oriented by a normal vector N, and s ∈ {1, . . . , n − 1}. Then<br />
∫<br />
)<br />
˜kn<br />
(∫S∩L (JÑ)dx dL s<br />
s<br />
L C s<br />
= vol(G C n−2,s−1) ω ( )<br />
2n−2 n −1 ( ∫<br />
)<br />
2sn − s − n<br />
k n (JN) + (2n − 1)M 1 (S) ,<br />
2s s n − s S<br />
where ˜k n (JÑ) the normal curvature of S ∩ L s <strong>in</strong> the direction JÑ, k n(JN) the normal curvature<br />
of JN <strong>in</strong> CK n (ɛ) and ω 2n−2 denotes the volume of the unit ball <strong>in</strong> the standard Euclidean<br />
<strong>space</strong> of dimension 2n − 2.<br />
Proof. Denote JÑ by e s.<br />
By Lemma 3.1.3 we have ˜k n (JÑ) = k n(e s )<br />
. Us<strong>in</strong>g equalities (3.6) and (3.8) we obta<strong>in</strong><br />
〈JN, e s 〉<br />
I =<br />
∫<br />
=<br />
∫L s<br />
∫<br />
S<br />
S∩L s<br />
∫ ∫RP 2n−2<br />
˜ kn (JÑ)dxdL s<br />
G C n−2,s−1<br />
ω is<br />
〈JN, e s 〉 2s<br />
(1 − 〈JN, e s 〉 2 ) s−1 k n (e s )<br />
〈JN, e s 〉 dV de sdp<br />
As the <strong>in</strong>tegral over G C n−2,s−1 is <strong>in</strong>dependent of V , it follows<br />
∫<br />
I = vol(G C 〈JN, e s 〉<br />
n−2,s−1)<br />
S<br />
∫RP 2s−1<br />
2n−2 (1 − 〈JN, e s 〉 2 ) s−1 k n(e s )de s dp. (3.9)<br />
In order to compute the <strong>in</strong>tegral over RP 2n−2 , we use polar coord<strong>in</strong>ates and express the<br />
normal curvature of e s <strong>in</strong> terms of the pr<strong>in</strong>cipal curvatures of T p S.<br />
That is, if {f 1 , . . . , f 2n−1 } is an orthonormal basis of pr<strong>in</strong>cipal directions of T p S then<br />
e s = ∑ 2n−1<br />
j=1 〈e s, f j 〉f j , and<br />
2n−1<br />
∑<br />
2n−1<br />
∑<br />
k n (e s ) = 〈e s , f j 〉 2 k n (f j ) = 〈e s , f j 〉 2 k j .<br />
j=1<br />
On the other hand, we consider a polar coord<strong>in</strong>ates system θ 1 , θ 2 with respect to JN<br />
def<strong>in</strong>ed by<br />
|〈JN, e s 〉| = cos θ 1 , (3.10)<br />
j=1
3.3 Mean curvature <strong>in</strong>tegral 53<br />
and us<strong>in</strong>g spherical trigonometry,<br />
〈e s , f j 〉 = cos θ 1 cos(JN, f j ) + s<strong>in</strong> θ 1 s<strong>in</strong>(JN, f j ) cos(e s , JN, f j )<br />
= cos θ 1 cos α j + s<strong>in</strong> θ 1 s<strong>in</strong> α j cos θ 2 (3.11)<br />
where cos(e s , JN, f j ) denotes the cos<strong>in</strong>e of the spherical angle with vertex JN. Note that α j<br />
are constants when the po<strong>in</strong>t is fixed.<br />
Then, from the relations<br />
Γ(s)Γ(n − s)<br />
Γ(n + 1)<br />
=<br />
)<br />
1 n −1<br />
and<br />
s(n − s)(<br />
s<br />
Γ(s)Γ(n − s + 1)<br />
Γ(n + 1)<br />
= 1 s<br />
( n<br />
s<br />
) −1<br />
,<br />
we get<br />
∫RP 2n−2 |〈JN, e s 〉| 2s−1<br />
=<br />
+<br />
=<br />
2n−1<br />
∑<br />
j=1<br />
∫ π ∫S 2n−4<br />
2n−1<br />
∑<br />
j=1<br />
= ω 2n−2<br />
2s<br />
(1 − 〈JN, e s 〉 2 ) s−1 k n(e s )de s<br />
k j<br />
( ∫ S 2n−3 ∫ π/2<br />
0<br />
0<br />
cos 2s−1 θ 1<br />
s<strong>in</strong> 2s−2 θ 1<br />
cos 2 θ 1 cos 2 α j s<strong>in</strong> 2n−3 θ 1 dθ 1 dS 2n−3 +<br />
cos 2 θ 2 s<strong>in</strong> 2n−4 θ 2<br />
∫ π/2<br />
(<br />
k j O 2n−3 cos 2 (2sn − s − n)Γ(s)Γ(n − s)<br />
α j +<br />
4(n − 1)Γ(n + 1)<br />
( n −1 2n−1 ∑<br />
( )<br />
2sn − n − s<br />
k j cos<br />
s) 2 α j + 1 .<br />
n − s<br />
j=1<br />
Integrat<strong>in</strong>g over S and us<strong>in</strong>g<br />
we obta<strong>in</strong> the stated result.<br />
0<br />
k n (JN) =<br />
2n−1<br />
∑<br />
j=1<br />
cos 2s−1 θ<br />
)<br />
1<br />
s<strong>in</strong> 2s−2 s<strong>in</strong> 2 θ 1 s<strong>in</strong> 2 α j s<strong>in</strong> 2n−3 θ 1 dθ 1 dθ 2 dS 2n−4 + 0<br />
θ 1<br />
)<br />
k j 〈JN, f j 〉 2 =<br />
Γ(s)Γ(n − s + 1)<br />
4(n − 1)Γ(n + 1)<br />
2n−1<br />
∑<br />
j=1<br />
k j cos 2 α j<br />
Theorem 3.3.2. Let S ⊂ CK n (ɛ) be a compact (possibly with boundary) hypersurface of class<br />
C 2 oriented by a normal vector N, and let s ∈ {1, ..., n − 1}. Then<br />
∫<br />
L C s<br />
M (s)<br />
1 (S∩L s)dL s = ω 2n−2vol(G C n−2,s−1 ) ( n −1 (<br />
(2n − 1)<br />
2s(2s − 1) s) 2ns − n − s ∫<br />
M 1 (S) +<br />
n − s<br />
where k n (JN) denotes the normal curvature <strong>in</strong> the direction JN ∈ T S.<br />
L C s<br />
S<br />
)<br />
k n (JN)<br />
Proof. By Proposition 3.2.2 and Lemma 3.1.4 we have<br />
∫<br />
M (s)<br />
1 (S ∩ L s)dL s = 1 ∫ ∫<br />
|〈JN, e s 〉|<br />
2s − 1<br />
∫RP 2s−1<br />
2n−2 (1 − 〈JN, e s 〉 2 ) s−1 σ 1(p; e s , V )dV de s dp<br />
= 1<br />
2s − 1<br />
∫<br />
S<br />
= vol(GC n−2,s−1 )<br />
2s − 1<br />
S<br />
S<br />
G C n−2,s−1<br />
|〈JN, e s 〉|<br />
∫RP 2s−1 ∫<br />
2n−2 (1 − 〈JN, e s 〉 2 ) s−1 (tr(II| V ) + II(e s , e s ))dV de s dp<br />
G C n−2,s−1<br />
∫<br />
|〈JN, e s 〉|<br />
∫RP 2s−1 ( )<br />
s − 1<br />
2n−2 (1 − 〈JN, e s 〉 2 ) s−1 n − 2 tr(II| E) + k n (e s ) de s dp,<br />
where E = 〈N, JN, e s , Je s 〉 ⊥ .
54 Average of the mean curvature <strong>in</strong>tegral<br />
Note that if s = 1 then dim V = 0. Although the <strong>in</strong>tegral ∫ G<br />
tr(II|V )dV has no<br />
C<br />
n−2,s−1<br />
sense, last equality above rema<strong>in</strong>s true s<strong>in</strong>ce s−1<br />
n−2 trII| E = 0.<br />
We shall study the follow<strong>in</strong>g <strong>in</strong>tegrals<br />
J E = vol(GC n−2,s−1 )<br />
2s − 1<br />
J s = vol(GC n−2,s−1 )<br />
2s − 1<br />
∫<br />
s − 1<br />
n − 2<br />
∫<br />
S<br />
S<br />
∫RP 2n−2 |〈JN, e s 〉| 2s−1<br />
(1 − 〈JN, e s 〉 2 ) s−1 tr(II| E)de s dp<br />
∫RP 2n−2 |〈JN, e s 〉| 2s−1<br />
(1 − 〈JN, e s 〉 2 ) s−1 k n(e s )de s dp.<br />
The second <strong>in</strong>tegral is the same (except for a constant factor) as the <strong>in</strong>tegral (3.9) <strong>in</strong> Proposition<br />
3.3.1. Thus, we know<br />
J s = O 2n−3vol(G C n−2,s−1 ) ( ) n −1 ( ∫<br />
)<br />
2sn − s − n<br />
k n (JN) + (2n − 1)M 1 (S) .<br />
4s(2s − 1)(n − 1) s n − s S<br />
In order to study the <strong>in</strong>tegral J E , we shall use polar coord<strong>in</strong>ates <strong>in</strong> the same way and with<br />
the same notation as <strong>in</strong> the proof of Theorem 3.3.1. Let {e 1 , Je 1 , . . . , e s−1 , Je s−1 } be a J-basis<br />
of E ∩ T p L s and let {e s+1 , Je s+1 , . . . , e n−1 , Je n−1 } be a J-basis of E ∩ (T p L s ) ⊥ . With respect<br />
to this orthonormal basis of E<br />
tr(II| E ) =<br />
n−1<br />
∑<br />
i=1,i≠s<br />
(k n (e i ) + k n (Je i )).<br />
If we denote by {f 1 , . . . , f 2n−1 } a basis of pr<strong>in</strong>cipal directions of S at p, we obta<strong>in</strong><br />
k n (e i ) =<br />
j=1<br />
2n−1<br />
∑<br />
j=1<br />
k n (f j )〈e i , f j 〉 2 =<br />
2n−1<br />
∑<br />
and us<strong>in</strong>g polar coord<strong>in</strong>ates with respect to JN, we get<br />
⎛<br />
⎞<br />
2n−1<br />
∑<br />
n−1<br />
∑<br />
tr(II| E ) = k j<br />
⎝ (〈e i , f j 〉 2 + 〈Je i , f j 〉 2 ) ⎠<br />
=<br />
2n−1<br />
∑<br />
j=1<br />
k j<br />
⎛<br />
⎝<br />
i=1,i≠s<br />
n−1<br />
∑<br />
i=1,i≠s<br />
j=1<br />
k j 〈e i , f j 〉 2 ,<br />
⎞<br />
(cos 2 (e i , JN, f j ) + cos 2 (Je i , JN, f j ) ⎠ .<br />
We denote by (u, v, w) the spherical angle with vertex v and sides on u and w, cos φ ij =<br />
cos(e i , JN, f j ) and cos φ ij<br />
= cos(Je i , JN, f j ). Then, to study the <strong>in</strong>tegral J E we have to deal<br />
with the follow<strong>in</strong>g <strong>in</strong>tegral<br />
∫S 2n−3 ∫ π/2<br />
0<br />
cos 2s−1 θ 1<br />
s<strong>in</strong> 2s−2 θ 1<br />
s<strong>in</strong> 2 α j s<strong>in</strong> 2n−3 θ 1·<br />
· (cos 2 φ 1j + · · · + cos 2 φ s−1,j + cos 2 φ s+1,j + · · · + cos 2 φ n−1,j )dθ 1 dS 2n−3<br />
= s<strong>in</strong> 2 Γ(s)Γ(n − s)<br />
α j ·<br />
2Γ(n)<br />
∫<br />
· (cos 2 φ 1j + · · · + cos 2 φ s−1,j + cos 2 φ s+1,j + · · · + cos 2 φ n−1,j )dS 2n−3<br />
S 2n−3<br />
where θ 1 and α i are def<strong>in</strong>ed <strong>in</strong> (3.10) and (3.11).
3.3 Mean curvature <strong>in</strong>tegral 55<br />
Denote by ˜S 2n−1 the subset of S 2n−1 def<strong>in</strong>ed by all po<strong>in</strong>ts not <strong>in</strong> span{N, JN}. Consider<br />
the well-def<strong>in</strong>ed map<br />
Π : ˜S2n−1 −→ {N, JN} ⊥<br />
v ↦→ proj {N,JN} ⊥ (v)<br />
||proj {N,JN} ⊥(v)||<br />
Note that Π(e a ) = e a , with a ∈ {1, 1, . . . , s − 1, s − 1, s + 1, s + 1, . . . , n − 1, n − 1}<br />
Then, V ⊥ <strong>in</strong>side {N, JN} ⊥ is<br />
E ⊥ ∩ {N, JN} ⊥ = ({N, JN, e s , Je s } ⊥ ) ⊥ ∩ {N, JN} ⊥ = {Π(e s ), JΠ(e s )}. (3.12)<br />
As cos(φ aj ) = cos(e a , JN, f j ) denotes the cos<strong>in</strong>e of the spherical angle with vertex JN and<br />
po<strong>in</strong>ts <strong>in</strong> e a and f j , by def<strong>in</strong>ition, it co<strong>in</strong>cides with 〈Π(e a ), Π(f j )〉 = 〈e a , Π(f j )〉. Then, as Π(f j )<br />
is a unit vector conta<strong>in</strong>ed <strong>in</strong> the vector sub<strong>space</strong> with basis {e 1 , Je 1 , . . . , e s−1 , Je s−1 , Π(e s ), JΠ(e s )}<br />
it is satisfied<br />
1 = 〈Π(f j ), Π(f j )〉 2<br />
= 〈e 1 , Π(f j )〉 2 + · · · + 〈Je s−1 , Π(f j )〉 2 + 〈Π(e s ), Π(f j )〉 2 + 〈JΠ(e s ), Π(f j )〉 2<br />
.<br />
and we get<br />
∫<br />
(cos 2 φ 1j + · · · + cos 2 φ s−1,j + cos 2 φ s+1,j + ... + cos 2 φ n−1,j )dS<br />
S∫<br />
2n−3<br />
= (1 − 〈Π(e s ), Π(f j )〉 2 − 〈JΠ(e s ), Π(f j )〉 2 )dS.<br />
S 2n−3<br />
Now, we use polar coord<strong>in</strong>ates θ 2 , θ 3 with respect to Π(f j ) such that<br />
〈Π(e s ), Π(f j )〉 = cos θ 2 , θ 2 ∈ (0, π),<br />
and<br />
〈JΠ(e s ), Π(f j )〉 = s<strong>in</strong>(Π(e s ), Π(f j )) cos(Π(e s ), Π(f j ), JΠ(f j )) = s<strong>in</strong> θ 2 cos θ 3 , θ 3 ∈ (0, π).<br />
By Lemma 3.1.2 and the relation<br />
π<br />
O 2n−3 = O 2n−5<br />
n − 2<br />
we have<br />
∫ π ∫ π<br />
∫S 2n−5<br />
0<br />
= O 2n−3 −<br />
(1 − cos 2 θ 2 − s<strong>in</strong> 2 θ 2 cos 2 θ 3 ) s<strong>in</strong> 2n−4 θ 2 s<strong>in</strong> 2n−5 θ 3 dθ 3 dθ 2 dS 1<br />
0<br />
∫ π ∫S 2n−4<br />
0<br />
∫ π<br />
cos 2 θ 2 s<strong>in</strong> 2n−4 θ 2 −<br />
∫S 2n−5<br />
= O 2n−3 − O √ √<br />
2n−3<br />
πΓ(n − 2) πΓ(n −<br />
1<br />
2n − 2 − O 2n−52<br />
4Γ(n − 1 2 ) Γ(n)<br />
( π<br />
= O 2n−5<br />
n − 2 − π<br />
2(n − 1)(n − 2) − π<br />
2(n − 1)(n − 2)<br />
O 2n−5 π<br />
=<br />
(2n − 4)<br />
2(n − 1)(n − 2)<br />
= O 2n−5π<br />
2(n − 1) .<br />
0<br />
∫ π<br />
cos 2 θ 3 s<strong>in</strong> 2n−5 θ 3 s<strong>in</strong> 2n−2 θ 2<br />
2 )<br />
)<br />
0
56 Average of the mean curvature <strong>in</strong>tegral<br />
Thus,<br />
J E = O 2n−3vol(G C (<br />
n−2,s−1 )n(s − 1) n −1 (<br />
∫ )<br />
(2n − 1)M 1 (S) − k n (JN)<br />
2s(2s − 1)(n − s)(n − 1) s)<br />
S<br />
and add<strong>in</strong>g both expressions of J E and J s we get the result<br />
J E + J s = O 2n−3vol(G C n−2,s−1 ) ( ) n −1<br />
·<br />
4s(2s − 1)(n − 1) s<br />
(<br />
n(s − 1)<br />
· ((2n − 1) + (2n − 1)<br />
n − s )M 1(S) + ( 2sn − n − s<br />
)<br />
n(s − 1)<br />
−<br />
n − s n − s<br />
∫S<br />
) k n (JN)<br />
= O 2n−3vol(G C n−2,s−1 ) ( ) n −1<br />
·<br />
4s(2s − 1)(n − 1) s<br />
( ∫ )<br />
2n − 1<br />
·<br />
n − s (n − s + ns − n)M 2sn − n − s − ns + n<br />
1(S) + k n (JN)<br />
n − s<br />
S<br />
= O 2n−3vol(G C n−2,s−1 ) ( ) n −1 ( ∫ )<br />
s(2n − 1)(n − 1)<br />
M 1 (S) + k n (JN) .<br />
4s(2s − 1)(n − 1) s<br />
n − s<br />
S<br />
It is natural to ask which functionals we have to <strong>in</strong>tegrate over the <strong>space</strong> of <strong>complex</strong> s-planes<br />
to obta<strong>in</strong> the mean curvature <strong>in</strong>tegral of the <strong>in</strong>itial hypersurface.<br />
Theorem 3.3.3. Let S ⊂ CK n (ɛ) be a compact (possibly with boundary) oriented hypersurface<br />
of class C 2 . If we def<strong>in</strong>e<br />
ν(S) = (2ns − n − s)M 1 (S) − n − s ∫<br />
k n (JN)dx,<br />
2s − 1<br />
S<br />
then<br />
∫<br />
L C r<br />
ν(S ∩ L r )dL r = ω 2n−2vol(G C n−2,s−1 )2(n − 1)(2n − 1)(s − 1)<br />
M 1 (S).<br />
(2s − 1)<br />
Proof. The result follows straightforward from Theorems 3.3.1 and 3.3.2.<br />
3.4 Reproductive valuations<br />
Def<strong>in</strong>ition 3.4.1. Suppose given, for each s ∈ N, a valuation <strong>in</strong> CK n (ɛ), ϕ (s) . It is said that<br />
the collection {ϕ (s) } satisfies the reproductive property if for any regular doma<strong>in</strong> Ω,<br />
∫<br />
ϕ (s) (Ω ∩ L s )dL s = c n,s ϕ(Ω),<br />
for some constant c n,s depend<strong>in</strong>g on n and s.<br />
L C s<br />
Remark 3.4.2. Recall that mean curvature <strong>in</strong>tegral for regular doma<strong>in</strong>s extend to all K(C n ).<br />
Also ∫ ∂Ω k n(JN) extends to K(C n ) s<strong>in</strong>ce it co<strong>in</strong>cides with Γ 2n−2,n−1 (Ω) (cf. Example 2.4.20).<br />
As neither the mean curvature <strong>in</strong>tegral M (s)<br />
1 (∂Ω ∩ L s), nor the <strong>in</strong>tegral of the normal<br />
curvature, ∫ ∂Ω∩L s<br />
k n (JÑ)dx, satisfy the reproductive property, it is natural to ask whether<br />
there exists some l<strong>in</strong>ear comb<strong>in</strong>ation of these such that satisfies this property. We consider a<br />
l<strong>in</strong>ear comb<strong>in</strong>ation s<strong>in</strong>ce, <strong>in</strong> C n , they constitute a basis of Val U(n)<br />
n−2 (Cn ).
3.4 Reproductive valuations 57<br />
Theorem 3.4.3. Let Ω ⊂ CK n (ɛ) be a regular doma<strong>in</strong>, s ∈ {1, ..., n−1}. Consider the smooth<br />
valuations def<strong>in</strong>ed by<br />
∫<br />
ϕ 1 (Ω) = M 1 (∂Ω) − k n (JN)<br />
∂Ω<br />
and<br />
Then,<br />
and<br />
∫<br />
L C s<br />
∫<br />
ϕ 2 (Ω) = (2s − 1)(2n − 1)M 1 (∂Ω) + k n (JN).<br />
∂Ω<br />
ϕ 1 (Ω ∩ L s )dL s = ω 2n−2vol(G C ( )<br />
n−2,s−1 )(s − 1)(2n − 1) n −1<br />
ϕ 1 (Ω)<br />
(2s − 1)(n − s)<br />
s<br />
∫<br />
L C s<br />
ϕ 2 (Ω ∩ L s )dL s = ω 2n−2vol(G C n−2,s−1 )<br />
(2s − 1)<br />
( ) n − 2 −1<br />
ϕ 2 (Ω).<br />
s − 1<br />
In C n , each of ϕ 1 (Ω) and ϕ 2 (Ω) expands a 1-dimensional sub<strong>space</strong> of reproductive valuations<br />
of degree 2n − 2.<br />
Proof. Let<br />
∫<br />
ν(Ω) = aM 1 (∂Ω) + b k n (JN).<br />
∂Ω<br />
We look for relations between a and b to be ν a reproductive valuation, that is,<br />
∫<br />
ν(Ω ∩ L s )dL s = λν(Ω).<br />
L C s<br />
From Theorems 3.3.1 and 3.3.2 we have<br />
∫<br />
∫ (<br />
)<br />
ν(Ω ∩ L s )dL s = aM 1 (∂Ω ∩ L s ) + b k n<br />
∫∂Ω∩L (JÑ) s<br />
L C s<br />
L C s<br />
= ω 2n−2vol(G C n−2,s−1 ) ( ) n −1((<br />
b(2s − 1)(2sn − n − s)<br />
) ∫<br />
a + k n (JN)+<br />
2s(2s − 1) s<br />
n − s<br />
∂Ω<br />
+ ( a(2ns − n − s)<br />
+ b(2s − 1) ) )<br />
(2n − 1)M 1 (∂Ω) .<br />
n − s<br />
Thus, ν is reproductive if and only if for some λ ∈ R<br />
( )<br />
a(2ns − n − s)<br />
(2n − 1)<br />
+ b(2s − 1) = λa,<br />
n − s<br />
b(2s − 1)(2sn − n − s)<br />
a + = λb.<br />
n − s<br />
Solv<strong>in</strong>g this system we get two solutions, a = −b, λ =<br />
1)(2n − 1), λ =<br />
2n(n − 1)<br />
.<br />
n − s<br />
2s(s − 1)(2n − 1)<br />
n − s<br />
and a = b(2s −<br />
Remark 3.4.4. Last theorem gives all valuations <strong>in</strong> Val U(n)<br />
n−2 (Cn ) such that they satisfy the<br />
reproductive property. Why are these valuations reproductive Are they special <strong>in</strong> some<br />
sense It shall be <strong>in</strong>terest<strong>in</strong>g to know the answer and also to have a geometric <strong>in</strong>terpretation<br />
for these valuations.
58 Average of the mean curvature <strong>in</strong>tegral<br />
3.5 Relation with some valuations def<strong>in</strong>ed by Alesker<br />
In Section 2.3 we recalled the def<strong>in</strong>ition of valuations U k,p <strong>in</strong> C n . They constitute a basis for<br />
Val U(n) (C n ). From this basis it can be established the follow<strong>in</strong>g theorem by Alesker<br />
Theorem 3.5.1. (Theorem 3.1.2 [Ale03]) Let Ω be a regular doma<strong>in</strong> <strong>in</strong> C n . Let 0 < q <<br />
n, 0 < 2p < k < 2q. Then,<br />
∫<br />
L C q<br />
U k,p (Ω ∩ L q )dL q =<br />
where constants γ p depend only on n, q and p.<br />
[k/2]+n−q<br />
∑<br />
p=0<br />
γ p · U k+2(n−q),p (Ω),<br />
In the follow<strong>in</strong>g theorem we give the constants γ p with arbitrary n, q and k = 2q − 2.<br />
Theorem 3.5.2. Let Ω be a regular doma<strong>in</strong> and 0 < q < n. Then,<br />
∫<br />
U 2q−2,p (Ω ∩ L q )dL q = ω 2q−2ω 2n−2 vol(G C n−2,q−1 )vol(GC q−2,q−p−1 )<br />
L C (q − p)(2q − 2p − 1) ( )(<br />
n−2 q−2<br />
) ·<br />
q<br />
q−1 q−p−1<br />
( )<br />
(2n − 3)(n − 1)(n − q + p)<br />
·<br />
U 2n−2,1 (Ω) − (2n − 1)n(n − q + p − 1)U 2n−2,0 (Ω) .<br />
ω 2n−2<br />
First, we express ∫ ∂Ω k n(JN) (a translation <strong>in</strong>variant cont<strong>in</strong>uous valuation) <strong>in</strong> terms of<br />
{U k,p }.<br />
Proposition 3.5.3. Let Ω be a regular doma<strong>in</strong> <strong>in</strong> C n . Then<br />
∫<br />
k n (JN)dp = n(2n − 3)(2n − 2)2 ω 2<br />
U 2n−2,1 (Ω) − 2n(2n − 1)(2n 2 − 4n + 1)ω 2 U 2n−2,0 (Ω).<br />
ω 2n−2<br />
∂Ω<br />
Proof. From the relations among valuations {U k,p } and mean curvature <strong>in</strong>tegrals (see (2.3))<br />
we have<br />
U 2n−2,0 (Ω) = 1 M 1 (∂Ω) (3.13)<br />
2nω 2<br />
and<br />
∫<br />
1<br />
U 2n−2,1 (Ω) =<br />
M 1 (∂Ω ∩ L n−1 )dL n−1 .<br />
(2n − 2)ω 2<br />
L C n−1<br />
Us<strong>in</strong>g Proposition 3.3.2 with s = n − 1, we get the result.<br />
Proof of the theorem 3.5.2. From the def<strong>in</strong>ition of U k,p we have<br />
∫<br />
1<br />
U k,p (Ω) =<br />
M k−2p (Ω ∩ L n−p )dL n−p<br />
2(n − p)ω 2n−k<br />
and from Theorem 3.3.2<br />
∫<br />
1<br />
U 2q−2,p (Ω ∩ L q ) =<br />
2(q − p)ω 2<br />
L C q−p<br />
G C n,n−p<br />
M 1 ((∂Ω ∩ L q ) ∩ L q−p )dL q−p<br />
) −1·<br />
O 2q−3 vol(G C q−2,q−p−1<br />
=<br />
) ( q<br />
8(q − p) 2 (q − 1)(2q − 2p − 1)ω 2 q − p<br />
(<br />
)<br />
2q(q − p) − 2q + p<br />
· (2q − 1) M 1 (∂Ω ∩ L q ) + k n<br />
p<br />
∫∂Ω∩L (JÑ) q<br />
where Ñ denotes the normal <strong>in</strong>ward vector field to ∂Ω ∩ L q as a hypersurface <strong>in</strong> L q . Us<strong>in</strong>g<br />
aga<strong>in</strong> Theorems 3.3.1 and 3.3.2, we express the <strong>in</strong>tegrals over L C q as an <strong>in</strong>tegral oer ∂Ω. F<strong>in</strong>ally,<br />
from the relation <strong>in</strong> Proposition 3.5.3 and (3.13) we get the result.
3.6 Example: sphere <strong>in</strong> CK 3 (ɛ) 59<br />
3.6 Example: sphere <strong>in</strong> CK 3 (ɛ)<br />
In this section we check Theorem 3.3.2 <strong>in</strong> the case of a sphere of radius R <strong>in</strong> CK 3 (ɛ).<br />
On page 18 we give an expression for the pr<strong>in</strong>cipal curvatures of a sphere of radius R <strong>in</strong><br />
CK n (ɛ). Us<strong>in</strong>g this expression we get<br />
∫<br />
and<br />
k n (JN) = 2 cot ɛ (2R)V (∂B R ) = 2 cos2 ɛ(R) + s<strong>in</strong> 2 ɛ(R)<br />
∂B R<br />
2 s<strong>in</strong> ɛ (R) cos ɛ (R)<br />
π 3<br />
3! 6 s<strong>in</strong>5 ɛ(R) cos ɛ (R)<br />
= π 3 (cos 2 ɛ(R) + s<strong>in</strong> 2 ɛ(R)) s<strong>in</strong> 4 ɛ(R) = π 3 (2 s<strong>in</strong> 6 ɛ(R) + s<strong>in</strong> 4 ɛ(R))<br />
= π 3 (2 cos 6 ɛ(R) − 5 cos 4 ɛ(R) + 4 cos 2 ɛ(R) − 1)<br />
M 1 (∂B R ) = π3 6<br />
3!4 (5 s<strong>in</strong>4 ɛ(R) cos 2 ɛ(R) + s<strong>in</strong> 6 ɛ(R)) = π3<br />
4 (6 s<strong>in</strong>6 ɛ(R) + 5 s<strong>in</strong> 4 ɛ(R))<br />
= π 3 ( 6 5 cos6 ɛ(R) − 13<br />
5 cos4 ɛ(R) + 8 5 cos2 ɛ(R) − 1 5 ).<br />
Thus, the right hand side of Theorem 3.3.2 is<br />
O 3 π 3<br />
144 ((42 + 2) cos6 ɛ(R) − (13 · 7 + 5) cos 4 ɛ(R) + (56 + 4) cos 2 ɛ(R) − (7 + 1))<br />
( 11<br />
= 2π 5 36 cos6 ɛ(R) − 2 3 cos4 ɛ(R) + 5<br />
12 cos2 ɛ(R) − 1 )<br />
.<br />
18<br />
The left hand side of Theorem 3.3.2 is<br />
∫<br />
M (2)<br />
1 (∂B R ∩ L 2 )dL 2 .<br />
L C 2<br />
Let us compute first M (2)<br />
1 (∂B R ∩ L 2 ), for a fixed <strong>complex</strong> 2-plane, L 2 . Recall that the <strong>in</strong>tersection<br />
between a sphere and L 2 is a sphere <strong>in</strong>side L 2 with radius r satisfy<strong>in</strong>g cos ɛ (R) =<br />
cos ɛ (r) cos ɛ (ρ) where ρ is the distance from the orig<strong>in</strong> of the sphere B R to the plane L 2 (cf.<br />
[Gol99, Lemma 3.2.13]).<br />
Thus,<br />
∫<br />
L C 2<br />
= 1 12<br />
M (2)<br />
1 (∂B R ∩ L 2 ) = 1 ∫<br />
(2 cot ɛ (r) + 2 cot ɛ (2r))<br />
3 ∂B R<br />
M (2)<br />
1 (∂B R ∩ L 2 )dL 2<br />
∫<br />
G C 3,1<br />
∫S 1 ∫ R<br />
= 2πV (GC 3,1 )2<br />
12<br />
= πV (GC 3,1 )<br />
3<br />
0<br />
∫ R<br />
0<br />
= 2 3 (cot ɛ(r) + cot ɛ (2r)) 4π2 s<strong>in</strong> 3<br />
2!<br />
ɛ(r) cos ɛ (r)<br />
= 1<br />
12 (4 cos4 ɛ(r) − 5 cos 2 ɛ(r) + 1)<br />
= 1 1<br />
12 cos 4 ɛ(R) (4 cos4 ɛ(R) − 5 cos 2 ɛ(R) cos 2 ɛ(ρ) + cos 4 ɛ(ρ)).<br />
cos 4 ɛ(ρ)<br />
cos 4 ɛ(ρ) (4 cos4 ɛ(R) − 5 cos 2 ɛ(R) cos 2 ɛ(ρ) + cos 4 ɛ(ρ))2 cos ɛ (ρ) s<strong>in</strong> ɛ (ρ)<br />
(4 cos 4 ɛ(R) − 5 cos 2 ɛ(R) cos 2 ɛ(ρ) + cos 4 ɛ(ρ))2 cos ɛ (ρ) s<strong>in</strong> ɛ (ρ)<br />
( 11<br />
12 cos6 ɛ(R) − 2 cos 4 ɛ(R) + 5 4 cos2 ɛ(R) − 1 6<br />
and we get the same result <strong>in</strong> both side of the expression <strong>in</strong> Theorem 3.3.2.<br />
)
Chapter 4<br />
Gauss-Bonnet Theorem and Crofton<br />
formulas for <strong>complex</strong> planes<br />
In this chapter we obta<strong>in</strong> an expression for the measure of <strong>complex</strong> r-planes <strong>in</strong>tersect<strong>in</strong>g a<br />
compact doma<strong>in</strong> <strong>in</strong> CK n (ɛ). That is, we give an expression of<br />
∫<br />
χ(Ω ∩ L r )dL r (4.1)<br />
L C r<br />
for a regular doma<strong>in</strong> Ω ⊂ CK n (ɛ) as a l<strong>in</strong>ear comb<strong>in</strong>ation of the so-called Hermitic <strong>in</strong>tr<strong>in</strong>sic<br />
volumes valuations <strong>in</strong> CK n (ɛ) (cf. Def<strong>in</strong>ition 2.4.11). The method we use consists on comput<strong>in</strong>g<br />
the variation, when the doma<strong>in</strong> moves along the flow <strong>in</strong>duced by a smooth vector field, of the<br />
measure of <strong>complex</strong> r-planes <strong>in</strong>tersect<strong>in</strong>g the convex doma<strong>in</strong>. From the theory of valuations on<br />
C n , we know that the expression is a l<strong>in</strong>ear comb<strong>in</strong>ation of certa<strong>in</strong> valuations. Thus, comput<strong>in</strong>g<br />
also the variation of these valuations and then compar<strong>in</strong>g both results we shall deduce the f<strong>in</strong>al<br />
expression.<br />
Us<strong>in</strong>g the same method we obta<strong>in</strong> an expression (cf. Theorem 4.4.1) for the Euler characteristic<br />
of a compact doma<strong>in</strong> <strong>in</strong> terms of its Gauss curvature of the boundary, its volume<br />
and others Hermitian <strong>in</strong>tr<strong>in</strong>sic volumes. This expression is analogous to the one obta<strong>in</strong>ed <strong>in</strong><br />
[San04, page 309] for real <strong>space</strong> <strong>forms</strong>.<br />
Relat<strong>in</strong>g these two results we shall obta<strong>in</strong> another expression for the Euler characteristic.<br />
This one <strong>in</strong>volves the measure of <strong>complex</strong> hyperplanes <strong>in</strong>tersect<strong>in</strong>g the regular doma<strong>in</strong> (cf.<br />
Theorem 4.4.5).<br />
4.1 Variation of the Hermitian <strong>in</strong>tr<strong>in</strong>sic volumes<br />
The study of the variation of a valuation when the doma<strong>in</strong> moves along the flow of a smooth<br />
vector field will be useful to deduce some properties of the valuation. In [BF08] it is given<br />
the variation of some valuations on C n and this variation is used to characterize monotone<br />
valuations. In this work, we give the variation of the Hermitian <strong>in</strong>tr<strong>in</strong>sic volumes (cf. Def<strong>in</strong>ition<br />
2.4.11) on CK n (ɛ) and we use it to deduce expression (4.1) <strong>in</strong> terms of these valuations.<br />
In order to obta<strong>in</strong> the variation of Hermitian <strong>in</strong>tr<strong>in</strong>sic volumes on CK n (ɛ) we follow the<br />
same method as <strong>in</strong> the proof of Corollary 2.6 <strong>in</strong> [BF08]. First, we recall the def<strong>in</strong>ition of the<br />
Rum<strong>in</strong> derivative, <strong>in</strong>troduced <strong>in</strong> [Rum94].<br />
Def<strong>in</strong>ition 4.1.1. Let µ ∈ Ω 2n−1 (S(CK n (ɛ))), let α be the contact form of S(CK n (ɛ)) and let<br />
α ∧ ξ ∈ Ω 2n−1 (S(CK n (ɛ))) be the unique (cf. [Rum94]) form such that d(µ + α ∧ ξ) is multiple<br />
of α. Then, the Rum<strong>in</strong> operator D is def<strong>in</strong>ed as<br />
Dµ := d(µ + α ∧ ξ).<br />
61
62 Gauss-Bonnet Theorem and Crofton formulas for <strong>complex</strong> planes<br />
Let us recall the def<strong>in</strong>ition of the Reeb vector field over a contact manifold.<br />
Def<strong>in</strong>ition 4.1.2. Let M be a contact manifold with contact form α. The Reeb vector field<br />
T is the only vector field over M such that<br />
{<br />
iT α = 1,<br />
L T α = 0.<br />
(4.2)<br />
If the contact manifold is the fiber tangent bundle of a Riemann manifold, then the Reeb<br />
vector field co<strong>in</strong>cides with the geodesic flow (cf. [Bla76, page 17]). This is the situation <strong>in</strong> this<br />
work, we consider the unit tangent bundle of CK n (ɛ) (cf. Lemma 1.3.7 and Remark 1.3.8).<br />
Note that the condition L T α = 0 is equivalent to i T dα = 0. Indeed,<br />
L T α = i T dα + d(i T α) = i T dα = dα(T ).<br />
The follow<strong>in</strong>g lemma conta<strong>in</strong>s the value of the contraction of T with α, β, γ and θ i def<strong>in</strong>ed<br />
<strong>in</strong> Section 2.4.2.<br />
Lemma 4.1.3. In S(CK n (ɛ)) it is satisfied<br />
i T α = 1, i T θ 1 = γ,<br />
i T θ 2 = β, i T β = i T γ = i T θ 0 = i T θ s = 0.<br />
Proof. The first equality is a characterization of the Reeb vector field. Moreover, i T θ s =<br />
−i t dα = di T α − L T α = 0.<br />
As T is the geodesic flow, it satisfies α i (T ) = β i (T ) = 0 and α 1i = β 1i = 0, i ∈ {2, ..., n}.<br />
We get the result us<strong>in</strong>g Def<strong>in</strong>ition (2.6) extended to CK n (ɛ).<br />
In [BF08] it is proved the follow<strong>in</strong>g lemma, which allow to calculate the variation of a<br />
valuation def<strong>in</strong>ed from an <strong>in</strong>variant smooth form. The result is proved <strong>in</strong> C n but the same<br />
rema<strong>in</strong>s true for ɛ ≠ 0, and for any Riemann manifold. Here we repeat the proof <strong>in</strong> detail for<br />
CK n (ɛ).<br />
Lemma 4.1.4 ([BF08] Lemma 2.5). Suppose that Ω ⊂ CK n (ɛ) is a regular doma<strong>in</strong>, N the<br />
outward unit vector field to ∂Ω, X is a smooth vector field on CK n (ɛ) with flow F t and µ a<br />
smooth valuation given by a (2n − 1)-form ρ <strong>in</strong> S(CK n (ɛ)). Then<br />
∫<br />
d<br />
dt∣ µ(F t (Ω)) = δ X µ(Ω) =<br />
t=0<br />
N(Ω)<br />
〈X, N〉 i T (Dβ)<br />
where T is the Reeb vector field of S(CK n (ɛ)) and Dρ is the Rum<strong>in</strong> operator of ρ.<br />
Proof. Let ˜X be a lift of X at S(CK n (ɛ)) such that it preserves α, i.e. L ˜X(α) = 0. Then, if<br />
˜F denotes the flow of ˜X
4.1 Variation of the Hermitian <strong>in</strong>tr<strong>in</strong>sic volumes 63<br />
δ X µ(Ω) = d ( ∫ )<br />
dt∣ β = d ( ∫ )<br />
∗ t=0 N(F t(Ω)) dt∣ ˜F t β<br />
t=0 N(Ω)<br />
∫<br />
=<br />
(1)<br />
=<br />
(3)<br />
=<br />
(4)<br />
=<br />
(5)<br />
=<br />
(6)<br />
=<br />
(7)<br />
=<br />
L ˜Xβ<br />
N(Ω)<br />
∫<br />
∫<br />
∫<br />
∫<br />
∫<br />
∫<br />
∫<br />
=<br />
N(Ω)<br />
N(Ω)<br />
N(Ω)<br />
N(Ω)<br />
N(Ω)<br />
N(Ω)<br />
N(Ω)<br />
∫<br />
(2)<br />
i ˜Xdβ + d(i ˜Xβ) = i ˜Xdβ<br />
N(Ω)<br />
i ˜XDβ − i ˜Xd(α ∧ η)<br />
i ˜XDβ<br />
i ˜X(α ∧ ρ)<br />
(i ˜Xα)ρ<br />
α( ˜X)i T (α ∧ ρ)<br />
〈X, N〉i T Dβ<br />
First, note that we can change the variation of F t (Ω) <strong>in</strong> the first <strong>in</strong>tegral for the Lie derivative<br />
of the <strong>in</strong>tegrated form s<strong>in</strong>ce N(F t (Ω)) = ˜F t (N(Ω)).<br />
For (1) and (2), we use the follow<strong>in</strong>g property of the Lie derivative, L ˜Xβ<br />
= i ˜Xdβ + d(i ˜Xβ),<br />
and that the second term is an exact form, thus the <strong>in</strong>tegral vanishes.<br />
Equality (3) follows directly from the def<strong>in</strong>ition of the Rum<strong>in</strong> operator.<br />
For (4) we use<br />
i ˜Xd(α ∧ η) = L ˜X(α ∧ η) − d(i ˜X(α ∧ η))<br />
and that the second term is an exact form. The first term can be rewritten as<br />
L ˜X(α ∧ η) = (L ˜Xα) ∧ η + α ∧ L ˜Xη,<br />
and so, its <strong>in</strong>tegral vanishes s<strong>in</strong>ce ˜X preserves α, which vanishes over the normal fiber bundle<br />
(cf. Lemma 1.3.12).<br />
As the Rum<strong>in</strong> operator is, by def<strong>in</strong>ition, a 2n-form multiple of α, and it is def<strong>in</strong>ed on the<br />
normal fiber bundle, we get (5).<br />
For (6), us<strong>in</strong>g the notion of contraction we get<br />
i ˜X(α ∧ ρ) = (i ˜Xα) ∧ ρ + α ∧ (i ˜Xρ).<br />
The second term vanishes over N(Ω).<br />
To get equality (7), we repeat the same argument as <strong>in</strong> (4) <strong>in</strong> order to obta<strong>in</strong> the form<br />
α ∧ ρ, which is the Rum<strong>in</strong> operator of β.<br />
F<strong>in</strong>ally, we recall the def<strong>in</strong>ition of α and that the <strong>in</strong>tegral is over the unit fiber normal<br />
bundle, so that the po<strong>in</strong>ts are (x, N) with x ∈ ∂Ω and N the unit normal vector on ∂Ω at<br />
x.<br />
The previous lemma allows us to compute the variation of any valuation given by a form,<br />
once we know its Rum<strong>in</strong> operator. In this chapter we give the variation of the Hermitian<br />
<strong>in</strong>tr<strong>in</strong>sic volumes <strong>in</strong> CK n (ɛ) (<strong>in</strong> [BF08] is given for ɛ = 0).<br />
In the follow<strong>in</strong>g lemma we give the derivative β k,q and γ k,q (def<strong>in</strong>ed <strong>in</strong> 2.4.6) us<strong>in</strong>g Lemma<br />
2.4.8. They will be used for the computation of the Rum<strong>in</strong> operator.
64 Gauss-Bonnet Theorem and Crofton formulas for <strong>complex</strong> planes<br />
Lemma 4.1.5. In CK n (ɛ)<br />
and<br />
dβ k,q = c n,k,q (θ n−k+q<br />
0 ∧ θ k−2q<br />
1 ∧ θ q 2 − ɛ(n − k + q)α ∧ β ∧ θn−k+q−1 0 ∧ θ k−2q<br />
1 ∧ θ q 2 )<br />
dγ k,q =c n,k,q (θ n−k+q<br />
0 ∧ θ k−2q<br />
1 ∧ θ q 2 − ɛθn−k+q−1 0 ∧ θ k−2q<br />
1 ∧ θ q+1<br />
2 − ɛα ∧ β ∧ θ n−k+q−1<br />
0 ∧ θ k−2q<br />
1 ∧ θ q 2<br />
(n − k + q − 1)<br />
− ɛ α ∧ γ ∧ θ n−k+q−2<br />
0 ∧ θ k−2q+1<br />
1 ∧ θ q 2<br />
2<br />
(n − k + q − 1)<br />
− ɛ β ∧ γ ∧ θ 01 ∧ θ n−k+q−2<br />
0 ∧ θ k−2q<br />
1 ∧ θ q 2<br />
2<br />
).<br />
From the previous lemma and follow<strong>in</strong>g the method used by [BF08] we can compute the<br />
variation of B k,q and Γ k,q <strong>in</strong> CK n (ɛ).<br />
Notation 4.1.6. We denote<br />
∫<br />
˜B k,q = ˜B k,q (Ω) :=<br />
∂Ω<br />
∫<br />
〈X, N〉β k,q and ˜Γk,q = ˜Γ k,q (Ω) :=<br />
∂Ω<br />
〈X, N〉γ k,q .<br />
Proposition 4.1.7. Let X be a smooth vector field def<strong>in</strong>ed on CK n (ɛ) and Ω ⊂ CK n (ɛ) be<br />
a regular doma<strong>in</strong>. The variation <strong>in</strong> CK n (ɛ) of valuations B k,q and Γ k,q with respect to X is<br />
given by<br />
and<br />
δ X B k,q (Ω) = 2c n,k,q (c −1<br />
n,k−1,q (k − 2q)2˜Γk−1,q − c −1<br />
n,k−1,q−1 (n + q − k)q˜Γ k−1,q−1<br />
+c −1<br />
n,k−1,q−1 (n + q − k + 1 2 )q ˜B k−1,q−1 − c −1<br />
n,k−1,q (k − 2q)(k − 2q − 1) ˜B k−1,q<br />
+ɛ(c −1<br />
n,k+1,q+1 (k − 2q)(k − 2q − 1) ˜B k+1,q+1 − c −1<br />
n,k+1,q (n − k + q)(q + 1 2 ) ˜B k+1,q ))<br />
δ X Γ k,q (Ω) = 2c n,k,q<br />
(<br />
c −1<br />
n,k−1,q (k − 2q)2˜Γk−1,q − c −1<br />
n,k−1,q−1 (n + q − k)q˜Γ k−1,q−1<br />
+c −1<br />
n,k−1,q−1 (n + q − k + 1 2 )q ˜B k−1,q−1 − c −1<br />
n,k−1,q (k − 2q)(k − 2q − 1) ˜B k−1,q<br />
+ɛ ( c −1<br />
n,k+1,q+1 2(k − 2q)(k − 2q − 1) ˜B k+1,q+1 − c −1<br />
n,k+1,q ((n − k + q)(2q + 3 2 ) − 1 2 (q + 1)) ˜B k+1,q<br />
−c −1<br />
n,k+1,q+1 (k − 2q)2˜Γk+1,q+1 + c −1<br />
n,k+1,q (n − k + q − 1)(q + 1)˜Γ k+1,q<br />
−ɛ(c −1<br />
n,k+3,q+2 (k − 2q)(k − 2q − 1) ˜B k+3,q+2 − c −1<br />
n,k+3,q+1 (n − k + q − 1)(q + 3 2 ) ˜B k+3,q+1 ) )) .<br />
Proof. We first study the valuation given by β k,q .<br />
Lemma 4.1.4 provides an expression for the variation of a smooth valuation. In order to<br />
use this lemma, it is enough to f<strong>in</strong>d i T Dβ k,q and i T γ k,q modulo α and dα s<strong>in</strong>ce the latter <strong>forms</strong><br />
vanish over N(Ω) (cf. Lemma 1.3.12).<br />
We will use the follow<strong>in</strong>g fact from the proof of Proposition 4.6 <strong>in</strong> [BF08]: for max{0, k −<br />
n} ≤ q ≤ k/2 < n there exists an <strong>in</strong>variant form ξ k,q ∈ Ω 2n−1 (S(C n )) such that<br />
and<br />
dα ∧ ξ k,q ≡ −θ n−k+q<br />
0 θ k−2q<br />
1 θ q 2 mod(α), (4.3)<br />
ξ k,q ≡ βγθ n+q−k−1<br />
0 θ k−2q−2<br />
1 θ q−1<br />
2 (4.4)<br />
∧ ( (n + q − k)qθ1 2 )<br />
− (k − 2q)(k − 2q − 1)θ 0 θ 2 mod(α, dα).
4.1 Variation of the Hermitian <strong>in</strong>tr<strong>in</strong>sic volumes 65<br />
In order to f<strong>in</strong>d δ X B k,q for general ɛ, we take a form ξ ɛ ∈ Ω 2n−1 (S(CK n (ɛ))) such that ξ ɛ (p,v) ≡<br />
ξ (p ′ ,v ′ ) when we identify T (p,v) S(CK n (ɛ)) and T (p ′ ,v ′ )C n , for every (p, v) ∈ S(CK n (ɛ)), (p ′ , v ′ ) ∈<br />
S(C n ). That is, as ξ ɛ is an <strong>in</strong>variant form, it can be expressed as a l<strong>in</strong>ear comb<strong>in</strong>ation of<br />
products with the <strong>forms</strong> α, β, θ 0 , θ 1 , θ 2 and θ s . We take this expression as a def<strong>in</strong>ition of ξ ɛ .<br />
From Lemma 4.1.5 and (4.3) we have that d(β k,q + c n,k,q α ∧ ξ ɛ ) ≡ 0 modulo α.<br />
By Lemma 2.4.8, the exterior differential of ξ ɛ is<br />
dξ ɛ ≡ θ n+q−k−1<br />
0 θ k−2q−2<br />
1 θ q−1<br />
2 ((n − k + q)qθ1 2 − (k − 2q)(k − 2q − 1)θ 0 θ 2 )<br />
∧ (γθ 1 − 2βθ 0 + 2ɛβθ 2 ) mod(α, dα)<br />
and the contraction of dβ k,q with respect to the vector field T , by Lemma 4.1.3, is<br />
i T dβ k,q ≡ c n,k,q θ n+q−k−1<br />
0 θ k−2q−1<br />
1 θ q−1<br />
2<br />
∧ ((k − 2q)γθ 0 θ 2 + qβθ 0 θ 1 − ɛ(n − k + q)βθ 1 θ 2 )<br />
mod(α).<br />
By substitut<strong>in</strong>g the last expressions <strong>in</strong> i T Dβ k,q ≡ i T dβ k,q − c n,k,q dξ ( mod α, dα), we get<br />
the result.<br />
The variation of Γ k,q with k ≠ 2q can be obta<strong>in</strong>ed us<strong>in</strong>g the relation among Γ k,q and B k,q<br />
given <strong>in</strong> Proposition 2.4.8 and the variation of B k,q .<br />
To compute δ X Γ 2q,q , note that dγ 2q,q has 3 terms not multiple of α (cf. Lemma 4.1.5). As<br />
before we take ξ ɛ 1 , ξɛ 2 ∈ Ω2n−1 (S(CK n (ɛ))) correspond<strong>in</strong>g to ξ 2q,q , and ξ 2q+2,q+1 respectively.<br />
Let us consider also<br />
ξ3 ɛ = n − q − 1 βγθ n−q−2<br />
0 θ q 2<br />
2<br />
. (4.5)<br />
Then the Rum<strong>in</strong> differential of γ 2q,q is given by Dγ 2q,q = d(γ 2q,q + c n,2q,q α ∧ (ξ1 ɛ − ɛξɛ 2 − ɛξɛ 3 )).<br />
Indeed, dα ∧ ξ1 ɛ cancels the first term of dγ 2q,q modulo α, and dα ∧ ξ2 ɛ cancels the second one.<br />
The third term is canceled exactly by dα ∧ ξ3 ɛ.<br />
Now, us<strong>in</strong>g Lemmas 4.1.5 and 4.1.3 we get<br />
i T dγ 2q,q ≡qβθ n−q<br />
0 θ q−1<br />
2 − ɛ(q + 2)βθ n−q−1<br />
and from (4.4) and (4.5)<br />
0 θ q 2 − ɛn − q − 1<br />
2<br />
γθ n−q−2<br />
0 θ 1 θ q 2 mod(α, dα),<br />
dξ ɛ 1 ≡ (n − q)qθ n−q−1<br />
0 θ q−1<br />
2 (γθ 1 − 2βθ 0 + 2ɛβθ 2 ) mod(α, dα).<br />
dξ ɛ 2 ≡ (n − q − 1)(q + 1)θ n−q−2<br />
0 θ q 2 (γθ 1 − 2βθ 0 + 2ɛβθ 2 ) mod(α, dα).<br />
dξ3 ɛ ≡ n − q − 1 θ n−q−2<br />
0 θ q 2<br />
2<br />
(γθ 1 − 2βθ 0 + 2ɛβθ 2 ) mod(α, dα).<br />
Plugg<strong>in</strong>g this <strong>in</strong>to i T Dγ 2q,q ≡ i T dγ 2q,q − c n,2q,q (dξ1 ɛ − ɛdξɛ 2 − ɛdξɛ 3 ) mod (α, dα) gives the result.<br />
Remark 4.1.8. For ɛ = 0 the variation of B k,q co<strong>in</strong>cides with the variation of Γ k,q and we get<br />
the result of Proposition 4.6 <strong>in</strong> [BF08].<br />
From the previous proposition we can obta<strong>in</strong> easily the variation of the Gauss curvature<br />
<strong>in</strong>tegral. We know that this variation vanishes <strong>in</strong> C n , for the Gauss-Bonnet theorem, but not<br />
<strong>in</strong> the other <strong>complex</strong> <strong>space</strong> <strong>forms</strong>.<br />
Corollary 4.1.9. In CK n (ɛ) the variation of the Gauss curvature <strong>in</strong>tegral is<br />
δ X M 2n−1 (∂Ω) = 2ɛω 2n−1 (2(n − 1)˜Γ 1,0 − (3n − 1) ˜B 1,0 + 3<br />
2π ɛ(2n − 1) ˜B 3,1 ).
66 Gauss-Bonnet Theorem and Crofton formulas for <strong>complex</strong> planes<br />
Proof. First, we relate the Gauss curvature <strong>in</strong>tegral with Γ 0,0 (Ω) from Example 2.4.20.1<br />
M 2n−1 (∂Ω) = 2c−1 n,0,0<br />
(n − 1)! Γ 0,0(Ω) = 2nω 2n Γ 0,0 (Ω). (4.6)<br />
Thus, from Proposition 4.1.7 and us<strong>in</strong>g that c −1<br />
n,1,0 = (n − 1)!ω 2n−1, c −1<br />
n,3,1 = (n − 2)!ω 2n−3 and<br />
ω 2n−1 = (2n − 1)ω 2n−3 /2π we obta<strong>in</strong> the result:<br />
δ X M 2n−1 (∂Ω) =<br />
2ɛ<br />
(n − 1)! (−c−1 n,1,0 (3n − 1) ˜B 1,0 + 2c −1<br />
n,1,0 (n − 1)˜Γ 1,0 + c −1<br />
n,3,1 3ɛ(n − 1) ˜B 3,1 )<br />
= 2ɛ(−ω 2n−1 (3n − 1) ˜B 1,0 + 2ω 2n−1 (n − 1)˜Γ 1,0 + 3ɛω 2n−3 ˜B3,1 )<br />
= 2ɛω 2n−1 (−(3n − 1) ˜B 1,0 + 2(n − 1)˜Γ 1,0 + 3<br />
2π ɛ(2n − 1) ˜B 3,1 ).<br />
4.2 Variation of the measure of <strong>complex</strong> r-planes <strong>in</strong>tersect<strong>in</strong>g<br />
a regular doma<strong>in</strong><br />
Proposition 4.2.1. Let Ω ⊂ CK n (ɛ) be a regular doma<strong>in</strong>, X a smooth vector field on CK n (ɛ)<br />
with flow φ t and Ω t = φ t (Ω). Then<br />
∫<br />
∫<br />
( ∫ )<br />
d<br />
dt∣ χ(Ω t ∩ L r )dL r = 〈∂φ/∂t, N〉 σ 2r (II| V )dV dx<br />
t=0 G C n,r (Dp)<br />
L C r<br />
∂Ω<br />
where N is the outward normal field, D is the tangent distribution to ∂Ω and orthogonal to<br />
JN and σ 2r (II| V ) denotes the 2r-th symmetric elementary function of II restricted to V ∈<br />
G C n−1,r (D p).<br />
Proof. The proof is similar to the one <strong>in</strong> [Sol06, Theorem 4] for real <strong>space</strong> <strong>forms</strong>.<br />
Denote G C n−1,r (D) = {(p, l) | l ⊆ T p∂Ω, dim R l = 2r and Jl = l} = ⋃ p∈∂Ω GC n−1,r (D p).<br />
For each V ∈ G C n−1,r (D p), we take the parallel translation V t of V along φ t (x). Recall that<br />
parallel translation preserves the <strong>complex</strong> structure (cf. [O’N83, page 326]). Then we project<br />
V t orthogonally to D φt(x), obta<strong>in</strong><strong>in</strong>g a <strong>complex</strong> r-plane V ′<br />
t (at least for small values of t). We<br />
def<strong>in</strong>e<br />
γ : G C n−1,r (D) × (−ɛ, ɛ) −→ LC r<br />
((x, V ), t) ↦→ exp φt(x) V t ′ .<br />
Proposition 3 <strong>in</strong> [Sol06] rema<strong>in</strong>s true, without change, <strong>in</strong> <strong>complex</strong> <strong>space</strong> <strong>forms</strong>. From this<br />
proposition and us<strong>in</strong>g a similar argument as <strong>in</strong> [Sol06, teorema 4] we get<br />
∫<br />
∫<br />
d<br />
1<br />
∑ ∂φ<br />
dt∣ χ(Ω t ∩ L r )dL r = lim<br />
sign〈<br />
t=0 L C h→0 h<br />
r<br />
G C n−1,r (D)×(0,h) ∂t , N〉 sign(σ 2r(II| V ))γ ∗ dL r<br />
∫<br />
= 〈 ∂φ<br />
G C n−1,r (D) ∂t , N〉 sign(σ 2r(II| V ))γ0(ι ∗ dφ∂t dL r )<br />
where the sum on the second <strong>in</strong>tegral runs over the tangencies of L r with the hypersurfaces<br />
∂Ω t with 0 < t < h.<br />
Consider a J-mov<strong>in</strong>g frame {g; g 1 , Jg 1 , ..., g n , Jg n } such that g((p, l), t) = φ(p, t), γ =<br />
〈g; g 1 , Jg 1 , ..., g r , Jg r 〉 ∩ CK n (ɛ) and Jg n ((p, l), t) = N t (outward unit vector to ∂Ω t at φ(p, t)).<br />
We may assume that the mov<strong>in</strong>g frame is def<strong>in</strong>ed <strong>in</strong> a neighborhood of L C r s<strong>in</strong>ce we are only<br />
<strong>in</strong>terested <strong>in</strong> regular po<strong>in</strong>ts of γ.<br />
(4.7)
4.2 Variation of the measure of <strong>complex</strong> r-planes <strong>in</strong>tersect<strong>in</strong>g a doma<strong>in</strong> 67<br />
Considerem the curve L r (t) given by the parallel translation of L r along the geodesic<br />
given by N, the outward normal vector to ∂Ω 0 . Recall that parallel translations preserves the<br />
<strong>complex</strong> structure (cf. [O’N83, page 326]). If P ∈ T Lr L C r denotes the tangent vector to L r (t)<br />
at t = 0, then<br />
ω i (P ) = 〈dg(P ), g i 〉 = 〈 d dt g(L r(t)), g i 〉 = 0, i ∈ {r + 1, r + 1, ..., n − 1, n − 1},<br />
ω n (P ) = 〈dg(P ), N〉 = 1, (4.8)<br />
ω ij (P ) = 〈∇g i (P ), g j 〉 = 〈 D dt g i(L r (t)), g j 〉 = 0,<br />
j ∈ {r + 1, r + 1, ..., n, n}, i ∈ {1, 1, ..., r, r}.<br />
The measure of <strong>complex</strong> r-planes <strong>in</strong> CK n (ɛ) is (cf. Proposition 1.5.5)<br />
From (4.8), we get<br />
dL r =<br />
∣<br />
n∧<br />
i=r+1<br />
ω i ∧ ω i<br />
∧<br />
i=1,...,r<br />
j=r+1,...,n<br />
dL r = |ω n |ι P dL r<br />
s<strong>in</strong>ce ι P dL r = | ∧ n−1<br />
h=r+1 ω h ∧ ω h ∧ ω n<br />
∧<br />
ωij |.Thus,<br />
with<br />
and we get<br />
ω ij ω ij<br />
∣ ∣∣∣∣∣∣∣<br />
.<br />
ι dγ∂t dL r = |ω n (dγ∂t)|ι P dL r + |ω n |ι dγ∂t ι P dL r<br />
ω n (dγ∂t) = 〈dg(dγ∂t), N〉 = 〈 ∂φ<br />
∂t , N〉,<br />
γ ∗ 0(ω n )(v) = 〈dg(dγ 0 (v)), N〉 = 0 ∀v ∈ T (p,l) G C n−1,r(T p ∂Ω 0 ),<br />
γ ∗ 0(ι dγ∂t dL r ) = |〈 ∂φ<br />
∂t , N〉|γ∗ 0(ι P dL r ).<br />
F<strong>in</strong>ally, us<strong>in</strong>g that ψ ∗ 0 (ι P dL r ) = |σ 2r (II| V )|dV dx, we get the result.<br />
Remark 4.2.2. The <strong>in</strong>tegral<br />
∫<br />
G C n,r<br />
σ 2r (II| V )dV (4.9)<br />
seems difficult to compute directly. However, we will f<strong>in</strong>d it by an <strong>in</strong>direct method. Recall that<br />
the analogous <strong>in</strong>tegral <strong>in</strong> real <strong>space</strong> <strong>forms</strong> is a multiple of an elementary symmetric function<br />
of the pr<strong>in</strong>cipal curvatures.<br />
For r = n − 1, the <strong>in</strong>tegral (4.9) can be easily computed <strong>in</strong> CK n (ɛ).<br />
Corollary 4.2.3. Let Ω ⊂ CK n (ɛ) be a regular doma<strong>in</strong>, X a smooth vector field on CK n (ɛ)<br />
with flow φ t , and Ω t = φ t (Ω). Then,<br />
∫<br />
d<br />
dt∣ t=0<br />
L C n−1<br />
χ(L n−1 ∩ Ω t )dL n−1 = ω 2n−1 ˜B1,0 (Ω).
68 Gauss-Bonnet Theorem and Crofton formulas for <strong>complex</strong> planes<br />
Proof. From Proposition 4.2.1 we get the result s<strong>in</strong>ce there is just one <strong>complex</strong> hyperplane<br />
tangent to a po<strong>in</strong>t <strong>in</strong> ∂Ω. Thus,<br />
∫<br />
∫<br />
∫<br />
d<br />
dt∣ χ(Ω t ∩ L n−1 )dL n−1 = 〈∂φ/∂t, N〉 σ 2n−2 (II| V )dV dx<br />
t=0 L C n−1<br />
∂Ω 0 G C n−1,n−1<br />
∫<br />
= 〈∂φ/∂t, N〉σ 2n−2 (II| D )dx<br />
∂Ω 0<br />
= 〈∂φ/∂t, N〉<br />
∫∂Ω β ∧ θn−1 0<br />
0<br />
(n − 1)!<br />
= c−1 n,1,0<br />
(n − 1)! ˜B 1,0 (Ω)<br />
= ω 2n−1 ˜B1,0 (Ω).<br />
4.3 Measure of <strong>complex</strong> r-planes meet<strong>in</strong>g a regular doma<strong>in</strong><br />
4.3.1 In the standard Hermitian <strong>space</strong><br />
Us<strong>in</strong>g that the measure of <strong>complex</strong> r-planes <strong>in</strong> C n meet<strong>in</strong>g a regular doma<strong>in</strong> is a l<strong>in</strong>ear comb<strong>in</strong>ation<br />
of the Hermitian <strong>in</strong>tr<strong>in</strong>sic volumes, and Propositions 4.2.1 and 4.1.7 we f<strong>in</strong>d explicitly<br />
the coefficients of this l<strong>in</strong>ear comb<strong>in</strong>ation.<br />
Theorem 4.3.1. Let Ω ⊂ C n be a convex doma<strong>in</strong>, X a smooth vector field over C n , φ t the<br />
flow associated to X and Ω t = φ t (Ω). Then<br />
∫<br />
( )<br />
d<br />
n − 1 −1 ( ) n<br />
dt∣ χ(Ω t ∩ L r )dL r = vol(G C n−1,r)ω 2r+1 (r + 1)<br />
−1·<br />
t=0 L C r r<br />
r<br />
⎛<br />
⎞<br />
n−r−1<br />
∑<br />
( )<br />
· ⎝<br />
2n − 2r − 2q − 1 1<br />
n − r − q 4 ˜B n−r−q−1 2n−2r−1,q (Ω) ⎠ , (4.10)<br />
and<br />
∫<br />
L C r<br />
q=max{0,n−2r−1}<br />
( ) n − 1 −1 ( n<br />
χ(Ω ∩ L r )dL r = vol(G C n−1,r)ω 2r<br />
r r<br />
⎛<br />
· ⎝<br />
n−r<br />
∑<br />
q=max{0,n−2r}<br />
) −1·<br />
( )<br />
1 2n − 2r − 2q<br />
4 n−r−q n − r − q<br />
Proof. In order to simplify the follow<strong>in</strong>g computations, we consider<br />
and<br />
B k,q ′ = B′ k,q (Ω) := c−1<br />
n,k,q µ k,q(Ω),<br />
˜B ′ k,q = c−1 n,k,q ˜B k,q ,<br />
⎞<br />
µ 2n−2r,q (Ω) ⎠ . (4.11)<br />
Γ ′ 2q,q = Γ ′ 2q,q(Ω) := 2c −1<br />
n,2q,q µ 2q,q(Ω). (4.12)<br />
˜Γ′ k,q = 2c −1<br />
n,k,q ˜Γ k,q . (4.13)<br />
The functional ∫ L C r χ(Ω ∩ L r)dL r is a valuation on C n with degree of homogeneity 2n −<br />
2r. Thus, it can be expressed as a l<strong>in</strong>ear comb<strong>in</strong>ation of the elements of degree 2n − 2r,<br />
{µ 2n−2r,q | max{0, n − 2r} ≤ q ≤ n − r} (cf. Def<strong>in</strong>ition 2.4.11). Then, by Remark 2.4.12 and<br />
(2.8), we have<br />
∫<br />
L C r<br />
χ(Ω ∩ L r )dL r =<br />
n−r−1<br />
∑<br />
q=max{0,n−2r}<br />
C q B ′ 2n−2r,q + DΓ ′ 2n−2r,n−r (4.14)
4.3 Measure of <strong>complex</strong> r-planes meet<strong>in</strong>g a regular doma<strong>in</strong> 69<br />
for certa<strong>in</strong> constants C q , D which we wish to determ<strong>in</strong>e. This will be done by compar<strong>in</strong>g the<br />
variation of both sides of this equality.<br />
From here on we assume 2r < n. The case 2r ≥ n can be treated <strong>in</strong> the same way (cf.<br />
Remark 4.3.2).<br />
By Proposition 4.1.7, the variation of the right hand side of (4.14) is a l<strong>in</strong>ear comb<strong>in</strong>ation<br />
of the follow<strong>in</strong>g type<br />
n−r−1<br />
∑<br />
q=n−2r−1<br />
c q ˜B′ 2n−2r−1,q +<br />
n−r−1<br />
∑<br />
q=n−2r<br />
d q ˜Γ′ 2n−2r−1,q (4.15)<br />
where the coefficients c q and d q can be expressed <strong>in</strong> terms of a l<strong>in</strong>ear comb<strong>in</strong>ation with known<br />
coefficients of the variables C q and D, that still rema<strong>in</strong> unknown.<br />
The variation of the left hand side of (4.14), by Proposition 4.2.1 is<br />
∫<br />
∫<br />
∫<br />
d<br />
dt∣ χ(Ω t ∩ L r )dL r = 〈∂φ/∂t, N〉 σ 2r (II| V )dV dx. (4.16)<br />
t=0<br />
L C r<br />
∂Ω<br />
G C n−1,r<br />
From Lemma 2.4.18 when pull<strong>in</strong>g-back the form γ k,q from N(Ω) to ∂Ω, one gets a polynomial<br />
expression P k,q of degree 2n − k − 1 <strong>in</strong> the coefficients h ij of II with i, j ∈ {1, 2, 2, . . . , n, n}.<br />
Moreover, for each q the monomials <strong>in</strong> P k,q conta<strong>in</strong><strong>in</strong>g only entries of the form h ii conta<strong>in</strong> the<br />
factor h 11 = II(JN, JN) and do not appear <strong>in</strong> any other P k,q ′ with q ′ ≠ q. Therefore, every<br />
non-trivial l<strong>in</strong>ear comb<strong>in</strong>ation of {P k,q } q must conta<strong>in</strong> the variable h 11 . On the other hand,<br />
the <strong>in</strong>tegral ∫ G<br />
σ C 2r (II| V )dV is a polynomial of the second fundamental form II restricted<br />
n−1,r<br />
to the distribution D = 〈N, JN〉 ⊥ , hence a polynomial not <strong>in</strong>volv<strong>in</strong>g h 11 . Compar<strong>in</strong>g the<br />
expressions of (4.15) and (4.16), it follows that d q = 0 for all q ∈ {n − 2r, . . . , n − r − 1}.<br />
As c q and d q depend on C q and D, we will obta<strong>in</strong> the value of c q once we know the value<br />
of C q and D. We will get their value from the equalities {d q = 0}. Note that this gives r<br />
equations, s<strong>in</strong>ce q runs from n − 2r to n − r − 1 <strong>in</strong> (4.15). As for the unknowns, we need to<br />
f<strong>in</strong>d r constants C q plus the constant D <strong>in</strong> (4.14).<br />
We will get an extra equation by tak<strong>in</strong>g II| D = Id and equat<strong>in</strong>g (4.16) to (4.15). Then,<br />
for any pair (n, r) we have a compatible l<strong>in</strong>ear system s<strong>in</strong>ce constants <strong>in</strong> (4.14) exist. Next we<br />
f<strong>in</strong>d the solution.<br />
Let us relate explicitly the coefficients {c q } and {d q } <strong>in</strong> (4.15) with C q and D <strong>in</strong> (4.14).<br />
To simplify the range of the subscripts, we denote d n−r−a with a = 1, . . . , r and c n−r−a with<br />
a = 1, . . . , r + 1.<br />
Coefficient d n−r−1 . From the variation of B ′ k,q <strong>in</strong> Cn (Proposition 4.1.7), the coefficient of<br />
˜Γ ′ 2n−2r−1,n−r−1 comes from the variation of B′ 2n−2r,n−r−1 and Γ′ 2n−2r,n−r . Then,<br />
d n−r−1 = −2r(n − r)D + (2n − 2r − 2(n − r − 1)) 2 C n−r−1<br />
= 4C n−r−1 − 2r(n − r)D. (4.17)<br />
Coefficient d n−r−a , a = 2, . . . , r. The coefficient of ˜Γ ′ 2n−2r−1,n−r−a comes from the variation<br />
of B 2n−2r,n−r−a ′ and B′ 2n−2r,n−r−a+1 . Then,<br />
d n−r−a = (2n − 2r − 2(n − r − a)) 2 C n−r−a − (2r + n − r − a + 1 − n)(n − r − a + 1)C n−r−a+1<br />
= 4a 2 C n−r−a − (r − a + 1)(n − r − a + 1)C n−r−a+1 . (4.18)<br />
Coefficient c n−r−1 . The coefficient of ˜B ′ 2n−2r−1,n−r−1 comes from the variation of B′ 2n−2r,n−r−1<br />
and Γ ′ 2n−2r,n−r . Then,<br />
c n−r−1 = 4(r + 1/2)(n − r)D − 4C n−r−1<br />
= 2(2r + 1)(n − r)D − 4C n−r−1 . (4.19)
70 Gauss-Bonnet Theorem and Crofton formulas for <strong>complex</strong> planes<br />
Coefficient c n−r−a , a = 2, . . . , r − 2. The coefficient of ˜B′ 2n−2r−1,n−r−a comes from the<br />
variation of B ′ 2n−2r,n−r−a and B′ 2n−2r,n−r−a+1 . Then,<br />
c n−r−a = −2(2a)(2a − 1)C n−r−a + 2(r − a + 3/2)(n − r − a + 1)C n−r−a+1<br />
= −4a(2a − 1)C n−r−a + (2r − 2a + 3)(n − r − a + 1)C n−r−a+1 . (4.20)<br />
Coefficient c n−2r−1 . The coefficient of ˜B ′ 2n−2r−1,n−2r−1 comes from the variation of B′ 2n−2r,n−2r .<br />
Then,<br />
c n−2r−1 = (2r − 2(r + 1) + 3)(n − r − (r + 1) + 1)C n−2r<br />
= (n − 2r)C n−2r . (4.21)<br />
Now, we solve the l<strong>in</strong>ear system given by {d n−r−a = 0} where a ∈ {1, . . . , r}. From<br />
equations (4.17) and (4.18) the system is given by<br />
{ r(n − r)D = 2Cn−r−1<br />
4a 2 C n−r−a = (n − r − a + 1)(r − a + 1)C n−r−a+1 .<br />
Thus,<br />
(n − r − a + 1) · · · · · (n − r)(r − a + 1) · · · · r<br />
C n−r−a =<br />
2 · 4 a−1 a 2 (a − 1) 2 · · · · · 1 2 D<br />
(n − r)!r!<br />
=<br />
2 2a−1 (n − r − a)!(r − a)!a!a! D<br />
= D ( )( n − r r<br />
2 2a−1 . (4.22)<br />
a a)<br />
To obta<strong>in</strong> the value of D, we compute ∫ G<br />
σ C 2r (p)dV and β 2n−2r−1,n−r−a ′ <strong>in</strong> case II| D(p) =<br />
n−1,r<br />
λId for λ > 0, which occurs when Ω is a geodesic ball. On one hand, we have<br />
∫<br />
σ 2r (p)(λId| V )dV = λ 2r vol(G C n−1,r).<br />
G C n−1,r<br />
On the other hand, if II| D = λId, then the connection <strong>forms</strong> satisfy α 1i = λω i and β 1i = λω i .<br />
Thus, θ 1 = 2λθ 2 and θ 0 = λ 2 θ 2 and we obta<strong>in</strong><br />
β ′ 2n−2r−1,n−r−a(p) = λ 2r (β ∧ θ r−a+1<br />
0 ∧ θ 2a−2<br />
1 ∧ θ n−r−a<br />
2 )(p)<br />
= 2 2a−2 λ 2r (β ∧ θ n−1<br />
2 )(p) = 2 2a−2 λ 2r (n − 1)!.<br />
So, the equation<br />
∑r+1<br />
vol(G C n−1,r) = c n−r−a 2 2a−2 (n − 1)!<br />
a=1<br />
must be satisfied.
4.3 Measure of <strong>complex</strong> r-planes meet<strong>in</strong>g a regular doma<strong>in</strong> 71<br />
Substitut<strong>in</strong>g equations (4.19), (4.20) and (4.21) <strong>in</strong> the last equation gives<br />
vol(G C n−1,r )<br />
(n − 1)!<br />
= (2(2r + 1)(n − r)D − 4C n−r−1 )<br />
r∑<br />
+ 2 2a−2 ((2r − 2a + 3)(n − r + a + 1)C n−r−a+1 − 4a(2a − 1)C n−r−a )<br />
a=2<br />
Thus,<br />
+ 2 2r (n − 2r)C n−2r<br />
= 2(2r + 1)(n − r)D + 4((2r − 1)(n − r − 1) − 1)C n−r−1<br />
∑r−1<br />
+ (−2 2a−2 4a(2a − 1) + 2 2a (2r − 2a + 1)(n − r − a))C n−r−a<br />
a=2<br />
+ (2 2r (n − 2r) − 2 2r−2 4r(2r − 1))C n−2r<br />
r∑<br />
= 2(2r + 1)(n − r)D + 2 2a ((2r − 2a + 1)(n − r − a) − a(2a − 1))C n−r−a<br />
(<br />
(4.22)<br />
= D 2(n − r)!r!<br />
2 n!<br />
= D<br />
r!(n − r − 1)!<br />
r∑<br />
a=0<br />
a=1<br />
D = vol(GC n−1,r )<br />
2 n!<br />
C n−r−a = vol(GC n−1,r )<br />
4 a n!<br />
(2r − 2a + 1)(n − r − a) − a(2a − 1)<br />
(n − r − a)!(r − a)!a!a!<br />
( ) n − 1 −1<br />
,<br />
r<br />
( ) n − 1 −1 ( )( n − r r<br />
r a a)<br />
and, for 2r < n, we have<br />
∫<br />
r∑<br />
χ(Ω ∩ L r )dL r = C n−r−a B 2n−2r,n−r−a ′ + DΓ ′ 2n−2r,n−r<br />
L C r<br />
and<br />
∫<br />
d<br />
dt∣ t=0<br />
L C r<br />
a=1<br />
= vol(GC n−1,r )<br />
2 n!<br />
( ) (<br />
n − 1<br />
−1 r∑ ( n − r<br />
r<br />
a<br />
a=1<br />
χ(Ω t ∩ L r )dL r = (2(2r + 1)(n − r)D − 4C n−r−1 )B ′ 2n−2r−1,n−r−1<br />
+<br />
)<br />
)( r<br />
a)<br />
2 −2a+1 B ′ 2n−2r,n−r−a + Γ ′ 2n−2r,n−r<br />
r∑<br />
((2r − 2a + 3)(n − r + a + 1)C n−r−a+1 − 4a(2a − 1)C n−r−a )B 2n−2r−1,n−r−a<br />
′<br />
a=2<br />
+ (n − 2r)C n−2r B 2n−2r−1,n−2r−1<br />
′<br />
= vol(GC n−1,r ) ( ) (<br />
n − 1<br />
−1 ∑r+1<br />
( )( ) )<br />
n − r r + 1 a<br />
n! r<br />
a a 4 ˜B ′ a−1 2n−2r−1,n−r−a . (4.23)<br />
a=1<br />
F<strong>in</strong>ally, we use the relation <strong>in</strong> (4.12) and (2.8) to obta<strong>in</strong> the result.<br />
Remark 4.3.2. If 2r ≥ n, then formula (4.10) follows directly from the relations among the<br />
different bases of valuations on C n given <strong>in</strong> [BF08] and the follow<strong>in</strong>g relation <strong>in</strong> [Ale03]<br />
∫<br />
χ(Ω ∩ L r )dL r = 1 ∫<br />
M 2r−1 (∂Ω ∩ L r )dL r = cU<br />
O 2(n−r),n−r<br />
2r−1<br />
L C r<br />
L C r<br />
)
72 Gauss-Bonnet Theorem and Crofton formulas for <strong>complex</strong> planes<br />
for a certa<strong>in</strong> constant c com<strong>in</strong>g from the different normalizations <strong>in</strong> dL r .<br />
Corollary 4.3.3. Let Ω ⊂ CK n (ɛ) be a regular doma<strong>in</strong>, X a smooth vector field over CK n (ɛ),<br />
φ t the flow associated to X and Ω t = φ t (Ω). Then<br />
∫<br />
( )<br />
d<br />
n − 1 −1 ( ) n<br />
dt∣ χ(Ω t ∩ L r )dL r = vol(G C n−1,r)ω 2r+1 (r + 1)<br />
−1·<br />
t=0 L C r r<br />
r<br />
⎛<br />
⎞<br />
n−r−1<br />
∑<br />
( )<br />
· ⎝<br />
2n − 2r − 2q − 1 1<br />
n − r − q 4 ˜B n−r−q−1 2n−2r−1,q (Ω) ⎠ . (4.24)<br />
q=max{0,n−2r−1}<br />
Proof. Compar<strong>in</strong>g equation (4.10) and Proposition 4.2.1 <strong>in</strong> case ɛ = 0 shows that if Ω is a<br />
regular convex doma<strong>in</strong>, then<br />
∫<br />
)<br />
〈X, N〉 σ 2r (II|V )dV dx<br />
∂Ω<br />
( ∫<br />
G C n,r<br />
equals the right hand side of equation above. By tak<strong>in</strong>g a vector field X that vanishes outside<br />
an arbitrarily small neighborhood of a fixed x ∈ ∂Ω, we deduce the follow<strong>in</strong>g equality between<br />
<strong>forms</strong><br />
(∫ )<br />
σ 2r (II| V )dV dx = ω 2r+1<br />
)( n<br />
G C n−1,r r)vol(G C n−1,r)(r + 1)· (4.25)<br />
(Tx∂Ω)<br />
·<br />
n−r−1<br />
∑<br />
q=max{0,n−2r−1}<br />
( n−1<br />
r<br />
( 2n − 2r − 2q − 1<br />
n − r − q<br />
)<br />
cn,2n−2r−1,n−r−q<br />
4 n−r−q−1 β ∧ θ r−q+1<br />
0 ∧ θ 2q−2<br />
1 ∧ θ n−r−q<br />
2 .<br />
This equation can be written as P (II)dx = Q(II)dx where P and Q are polyomials with entries<br />
<strong>in</strong> the second fundamental form. These polynomials concide for any positive def<strong>in</strong>ed bil<strong>in</strong>eal<br />
form. Thus, P = Q and (4.25) holds for regular doma<strong>in</strong>s (not necessarily convex doma<strong>in</strong>s).<br />
Moreover, it is valid <strong>in</strong> CK n (ɛ) for any ɛ. Apply<strong>in</strong>g Proposition 4.2.1 we get the result.<br />
Corollary 4.3.4. Equation (4.11) holds for any regular doma<strong>in</strong> not necessarily convex.<br />
Proof. Consider Ω t = φ t (Ω) with φ t a given flow.<br />
From the last corollary, it is known the variation of the left hand side of (4.11).<br />
By Proposition 4.1.7, the variation of the right hand side is a l<strong>in</strong>ear comb<strong>in</strong>ation of<br />
{B k,q , Γ k,q }. By Theorem 4.3.1 this l<strong>in</strong>ear comb<strong>in</strong>ation co<strong>in</strong>cides with the right hand side<br />
of (4.24).<br />
Thus, the variation of both sides of (4.11) co<strong>in</strong>cides. So, the difference between both<br />
members of (4.11) is constant.<br />
Take φ t such that φ t (Ω) converges to a po<strong>in</strong>t for t → ∞. Both sides of (4.11) tend to zero<br />
when t → ∞, thus their difference vanishes for all t.<br />
4.3.2 In <strong>complex</strong> <strong>space</strong> <strong>forms</strong><br />
Theorem 4.3.5. Let Ω be a regular doma<strong>in</strong> <strong>in</strong> CK n (ɛ). Then<br />
∫<br />
( ) n − 1<br />
χ(Ω ∩ L r )dL r = vol(G C n−1,r)<br />
−1· (4.26)<br />
L C r<br />
r<br />
⎛<br />
⎞<br />
n−1<br />
∑<br />
( ) n −1 ∑k−1<br />
( )<br />
· ( ɛ k−(n−r) ω 2n−2k<br />
⎝<br />
1 2k − 2q<br />
k<br />
4 k−q µ 2k,q (Ω) + (k + r − n + 1)µ 2k,k (Ω) ⎠<br />
k − q<br />
k=n−r<br />
+ ɛ r (r + 1)vol(Ω)).<br />
q=max{0,2k−n}
4.3 Measure of <strong>complex</strong> r-planes meet<strong>in</strong>g a regular doma<strong>in</strong> 73<br />
Proof. We will show that both sides have the same variation δ X with respect to any vector<br />
field X. This implies the result: one can take a deformation Ω t of Ω such that Ω t converges<br />
to a po<strong>in</strong>t. Then both sides of (4.26) have the same derivative, and both vanish <strong>in</strong> the limit.<br />
The variation of the left hand side of (4.26) is given by Corollary 4.3.3. The variation of<br />
the right hand side can be computed by us<strong>in</strong>g Proposition 4.1.7, and δ X V = 2 ˜B 2n−1,n−1 . In<br />
order to simplify the computations we rewrite the right hand side of (4.26) as<br />
+<br />
n−1<br />
∑<br />
j=n−r<br />
By Proposition 3.8 we have<br />
+<br />
n−1<br />
∑<br />
j=n−r<br />
C r (Ω) := vol(GC n−1,r )<br />
n!<br />
⎛<br />
ɛ j−n+r ⎝ j − n + r + 1 Γ ′ 2j,j +<br />
2<br />
δ X C r (Ω) = vol(GC n−1,r)<br />
n!<br />
( ) n − 1 −1<br />
{ɛ r (r + 1)n!V<br />
r<br />
j−1<br />
∑<br />
q=max(0,2j−n)<br />
( )<br />
1 n − j j<br />
4 j−q j − q)(<br />
q<br />
B ′ 2j,q<br />
⎞<br />
⎠}.<br />
( ) −1 n − 1<br />
[ɛ r n(r + 1)δ x ˜B′ 2n−1,n−1 (4.27)<br />
r<br />
ɛ j−n+r j − n + r + 1 {−2(n − j)j˜Γ ′ 2j−1,j−1 + 2ɛ(n − j − 1)(j + 1)˜Γ ′ 2j+1,j<br />
2<br />
+4(n − j + 1 (<br />
2 )j ˜B j + 1<br />
2j−1,j−1 ′ + 4ɛ − (n − j)(2j + 3 )<br />
2<br />
2 ) ˜B 2j+1,j ′ + 4ɛ 2 (n − j − 1)(j + 3 2 ) ˜B 2j+3,j+1}]<br />
′<br />
+<br />
n−1<br />
∑<br />
∑j−1<br />
j=n−r q=max{0,2j−n}<br />
ɛ j−n+r ( ) n − j j<br />
4 j−q {(2j − 2q)<br />
j − q)( 2˜Γ′ q<br />
2j−1,q<br />
−(n + q − 2j)q˜Γ ′ 2j−1,q−1 + 2(n + q − 2j + 1 2 )q ˜B ′ 2j−1,q−1 − 2(2j − 2q)(2j − 2q − 1) ˜B ′ 2j−1,q<br />
+2ɛ(2j − 2q)(2j − 2q − 1) ˜B ′ 2j+1,q+1 − 2ɛ(n − 2j + q)(q + 1 2 ) ˜B ′ 2j+1,q}.<br />
We will show that the previous expression is <strong>in</strong>dependent of ɛ; i.e. all the terms conta<strong>in</strong><strong>in</strong>g ɛ<br />
cancel out. Hence, δ X C r (Ω) co<strong>in</strong>cides with (4.24) s<strong>in</strong>ce we know this happens for ɛ = 0. This<br />
will f<strong>in</strong>ish the proof.<br />
We concentrate first on the terms with ˜B k,q ′ . By putt<strong>in</strong>g together similar terms, the third<br />
l<strong>in</strong>e of (4.27) is<br />
n−1<br />
∑<br />
h=n−r+1<br />
ɛ h−n+r 2{(h − n + r + 1)(n − h + 1 2 )h + (h − n + r)(h 2 − (n − h + 1)(2h − 1 )) (4.28)<br />
2<br />
+(h − n + r + 1)(n − h + 1)(h − 1 2 )} ˜B ′ 2h−1,h−1<br />
−ɛ r {(r + 2)n − 1} ˜B ′ 2n−1,n−1 + (2r + 1)(n − r) ˜B ′ 2n−2r−1,n−r−1.<br />
By putt<strong>in</strong>g together similar terms, the double sum <strong>in</strong> (4.27) (forgett<strong>in</strong>g for the moment<br />
the terms with ˜Γ ′ k,q ) becomes<br />
n−1<br />
∑<br />
h−2<br />
∑<br />
h=n−r a=max(−1,2h−n−1)<br />
−<br />
n−1<br />
∑<br />
h−1<br />
∑<br />
h=n−r a=max(0,2h−n)<br />
ɛ h−n+r ( )( )<br />
n − h h<br />
4 h−a−1 2(n + a − 2h + 3 h − a − 1 a + 1<br />
2 )(a + 1) ˜B 2h−1,a<br />
′<br />
ɛ h−n+r ( )<br />
n − h h<br />
4 h−a 2(2h − 2a)(2h − 2a − 1)<br />
h − a)(<br />
a<br />
˜B 2h−1,a<br />
′
74 Gauss-Bonnet Theorem and Crofton formulas for <strong>complex</strong> planes<br />
−<br />
+<br />
n∑<br />
h−1<br />
∑<br />
h=n−r+1 a=max(1,2h−n−1)<br />
n∑<br />
h−2<br />
∑<br />
h=n−r+1 a=max(0,2h−n−2)<br />
ɛ h−n+r ( n − h + 1<br />
4 h−a h − a<br />
ɛ h−n+r ( n − h + 1<br />
4 h−a−1 h − a − 1<br />
)( ) h − 1<br />
2(2h − 2a)(2h − 2a − 1)<br />
a − 1<br />
˜B 2h−1,a<br />
′<br />
)( h − 1<br />
a<br />
)<br />
2(n − 2h + a + 2)(a + 1 2 ) ˜B ′ 2h−1,a.<br />
Note that the terms with a = −1 or a = 2h − n − 2 vanish, if they occur. Then, one checks<br />
that all the terms <strong>in</strong> the above expression cancel out except those with h = n − r, n, and those<br />
with a = h − 1. Clearly the terms correspond<strong>in</strong>g to h = n − r are <strong>in</strong>dependent of ɛ. The terms<br />
with h = n sum up ɛ r (n − 1) ˜B 2n−1,n−1 ′ , and together with the similar term appear<strong>in</strong>g <strong>in</strong> (4.28)<br />
cancel out the first term <strong>in</strong> (4.27). F<strong>in</strong>ally, the terms with a = h − 1 are cancelled with the<br />
sum <strong>in</strong> (4.28).<br />
With a similar but shorter analysis one checks that the multiples of ˜Γ ′ k,q<br />
cancel out completely.<br />
This shows that (4.27) is <strong>in</strong>dependent of ɛ, and f<strong>in</strong>ishes the proof.<br />
Remark 4.3.6. The coefficients of µ k,q and vol <strong>in</strong> (4.26) were found by solv<strong>in</strong>g a l<strong>in</strong>ear system<br />
of equations, which we write down <strong>in</strong> the appendix.<br />
4.4 Gauss-Bonnet formula <strong>in</strong> CK n (ɛ)<br />
Theorem 4.4.1. Let Ω be a regular doma<strong>in</strong> <strong>in</strong> CK n (ɛ). Then<br />
ω 2n χ(Ω) =(n + 1)ɛ n vol(Ω) + (4.29)<br />
⎛<br />
⎞<br />
n−1<br />
∑ (n − c)ω 2n−2c ɛ c<br />
+<br />
n ( ∑c−1<br />
( )<br />
) ⎝<br />
1 2c − 2q<br />
n−1<br />
4 c−q µ 2c,q (Ω) + (c + 1)µ 2c,c (Ω) ⎠ .<br />
c − q<br />
c<br />
c=0<br />
q=max{0,2c−n}<br />
Remark 4.4.2. For ɛ = 0 we have the Gauss-Bonnet formula <strong>in</strong> C n ∼ = R 2n , where it is known<br />
χ(Ω) = 1<br />
2nω 2n<br />
M 2n−1 (∂Ω) = µ 0,0 (Ω),<br />
which co<strong>in</strong>cides with the expression <strong>in</strong> the previous result.<br />
Here we prove the certa<strong>in</strong>ty of (4.29) but <strong>in</strong> the appendix we give a constructive proof of<br />
the result.<br />
Proof. We proceed analogously to the proof of Theorem 4.3.5. In fact, the same computations<br />
of the previous proof show (<strong>in</strong> case r = n) that the right hand side of (4.29) has null variation.<br />
For ɛ = 0 equation (4.29) is the well know Gauss-Bonnet formula <strong>in</strong> C n ∼ = R 2n . For ɛ ≠ 0,<br />
we take a smooth deformation of Ω to get a doma<strong>in</strong> conta<strong>in</strong>ed <strong>in</strong> a ball of radius r. Under<br />
this deformation, the right hand side of (4.29) rema<strong>in</strong>s constant. By tak<strong>in</strong>g r small enough,<br />
the difference between both sides can be made arbitarily small. Hence, they co<strong>in</strong>cide.<br />
Although <strong>in</strong> (4.29) does not appear the Gauss curvature, we can easily get the follow<strong>in</strong>g<br />
expression.<br />
Corollary 4.4.3. Let Ω ⊂ CK n (ɛ) be a regular doma<strong>in</strong>. Then,<br />
2nω 2n χ(Ω) =M 2n−1 (∂Ω) + 2n(n + 1)ɛ n vol(Ω)+<br />
⎛<br />
n−1<br />
∑<br />
( ) n −1<br />
+ 2nω 2n−2c ɛ c ⎝<br />
c<br />
c=1<br />
∑c−1<br />
q=max{0,2c−n}<br />
⎞<br />
( )<br />
1 2c − 2q<br />
4 c−q µ 2c,q (Ω) + (c + 1)µ 2c,c (Ω) ⎠ .<br />
c − q
4.4 Gauss-Bonnet formula <strong>in</strong> CK n (ɛ) 75<br />
Proof. Apply relation (4.6) <strong>in</strong> (4.29).<br />
Remarks 4.4.4. 1. The Gauss-Bonnet-Chern formula <strong>in</strong> <strong>space</strong>s of constant sectional curvature<br />
k and even dimension, for a regular doma<strong>in</strong> Ω is given by<br />
O 2m−1 χ(Ω) = M 2n−1 (∂Ω) + c n−3 M 2n−3 (∂Ω) + · · · + c 1 M 1 (∂Ω) + (|k|) n/2 vol(Ω),<br />
where c j are known and depend on the curvature k.<br />
Note that <strong>in</strong> the previous expression appear all mean curvature <strong>in</strong>tegrals with odd subscript<br />
and the volume. In formula (4.29) <strong>in</strong> CK n (ɛ), ɛ ≠ 0, also appear all the Hermitian<br />
<strong>in</strong>tr<strong>in</strong>sic volumes <strong>in</strong> CK n (ɛ) with the first subscript odd.<br />
2. In [Sol06] it is given an expression of the Gauss-Bonnet-Chern formula <strong>in</strong> <strong>space</strong> of constant<br />
sectional curvature k us<strong>in</strong>g the measure of planes of codimension 2 meet<strong>in</strong>g the<br />
doma<strong>in</strong>. The obta<strong>in</strong>ed formula for Ω ⊂ RK n (ɛ) is<br />
nω n χ(Ω) = M n−1 (∂Ω) +<br />
2k ∫<br />
χ(Ω ∩ L n−2 )dL n−2 . (4.30)<br />
ω n−1 L n−2<br />
A natural question is whether <strong>in</strong> <strong>complex</strong> <strong>space</strong> <strong>forms</strong>, there exists a similar expression<br />
relat<strong>in</strong>g the Gauss curvature <strong>in</strong>tegral with the Euler characteristic and the measure of<br />
some <strong>complex</strong> planes meet<strong>in</strong>g the doma<strong>in</strong><br />
or<br />
c 0 χ(Ω) = n−1<br />
∑<br />
∫<br />
M 2n−1 (∂Ω) + c q<br />
q=1<br />
c 0 χ(Ω) = n−1<br />
∑<br />
n−1<br />
∑<br />
∫<br />
M 2n−1 (∂Ω) + M 2q+1 (∂Ω) + c q<br />
q=0<br />
L C q<br />
χ(Ω ∩ L q )dL q<br />
q=1<br />
L C q<br />
χ(Ω ∩ L q )dL q .<br />
Tak<strong>in</strong>g variation on both sides <strong>in</strong> these expressions, we get that these expressions cannot<br />
hold <strong>in</strong> general (for n = 2 and n = 3 we can choose constants satisfy<strong>in</strong>g them). Anyway,<br />
<strong>in</strong> Theorem 4.4.5 we give a similar expression. Perhaps, if we knew a formula for the<br />
measure of totally real planes meet<strong>in</strong>g a doma<strong>in</strong> we could f<strong>in</strong>d a more similar expression.<br />
3. For n = 2, the Gauss-Bonnet-Chern formula was already known <strong>in</strong> CK n (ɛ). It was given<br />
<strong>in</strong> [Par02]. From Theorem 4.4.1 we get the same expression, which can be written as<br />
χ(Ω) = 1 ( ( )<br />
)<br />
1 1<br />
π 2 2 Γ′ 0,0 + ɛ<br />
4 B′ 2,0 + Γ ′ 2,1 + 6ɛ 2 vol(Ω) .<br />
This expression can also be stated as<br />
χ(Ω) = 1 (<br />
2π 2 M 3 (∂Ω) + 3ɛ ( ∫<br />
M 1 (∂Ω) +<br />
2<br />
∂Ω<br />
)<br />
)<br />
k n (JN) + 12ɛ 2 vol(Ω)<br />
(4.31)<br />
and<br />
(<br />
χ(Ω) = 1<br />
∫<br />
2π 2 M 3 (∂Ω) + 2ɛ χ(∂Ω ∩ L 1 )dL 1 + ɛ ∫<br />
)<br />
k n (JN) + 12ɛ 2 vol(Ω) .<br />
L C 2<br />
1<br />
∂Ω<br />
In the follow<strong>in</strong>g result we express the Euler characteristic <strong>in</strong> terms of the Gauss curvature<br />
<strong>in</strong>tegral, the volume, the measure of <strong>complex</strong> hyperplanes meet<strong>in</strong>g a doma<strong>in</strong> and the valuations<br />
µ 2c,c . This formula generalizes (4.30) <strong>in</strong> <strong>complex</strong> <strong>space</strong> <strong>forms</strong>.
76 Gauss-Bonnet Theorem and Crofton formulas for <strong>complex</strong> planes<br />
Theorem 4.4.5. Let Ω ⊂ CK n (ɛ) be a regular doma<strong>in</strong> CK n (ɛ). Then,<br />
∫<br />
ω 2n χ(Ω) = ɛ χ(Ω ∩ L n−1 )dL n−1 +<br />
L C n−1<br />
= 1<br />
∫<br />
2n M 2n−1(∂Ω) + ɛ<br />
L C n−1<br />
n∑<br />
c=0<br />
ɛ c ω 2n<br />
µ 2c,c (Ω)<br />
ω 2c<br />
χ(Ω ∩ L n−1 )dL n−1 +<br />
n∑<br />
c=1<br />
ɛ c ω 2n<br />
µ 2c,c (Ω).<br />
ω 2c<br />
Proof. From Theorems 4.3.5 and 4.4.1 we get the stated formula<br />
⎛<br />
⎞<br />
n−1<br />
∑ ɛ c n−1<br />
c! ∑<br />
( )<br />
χ(Ω) = ⎝<br />
1 2c − 2q<br />
π c<br />
4 c−q B 2c,q (Ω) + (c + 1)Γ 2c,c (Ω) ⎠ + ɛn (n + 1)!<br />
c − q<br />
π n vol(Ω)<br />
c=0 q=max{0,2c−n}<br />
⎛<br />
⎞<br />
n−1<br />
∑ ɛ c n−1<br />
c! ∑<br />
( )<br />
= Γ 0,0 (Ω)+ ⎝<br />
1 2c − 2q<br />
π c 4 c−q B 2c,c (Ω) + (c + 1)Γ 2c,c (Ω) ⎠+ ɛn (n + 1)!<br />
c − q<br />
π n vol(Ω)<br />
c=1 q=max{0,2c−n}<br />
⎛<br />
⎞<br />
= Γ 0,0 (Ω) + ɛ n! n−1<br />
∑ ɛ c−1 c!π n−c n−1<br />
∑<br />
( )<br />
⎝<br />
1 2c − 2q<br />
π n n!<br />
4 c−q B 2c,c (Ω) + cΓ 2c,c (Ω) ⎠ +<br />
c − q<br />
+ ɛ n! n−1<br />
∑<br />
π n<br />
c=1<br />
c=1<br />
= Γ 0,0 (Ω) + ɛ n!<br />
π n ∫<br />
ɛ c−1 c!π n−c<br />
n!<br />
L C n−1<br />
q=max{0,2c−n}<br />
Γ 2c,c (Ω) + ɛn (n + 1)!<br />
π n vol(Ω)<br />
n−1<br />
∑<br />
χ(Ω ∩ L n−1 )dL n−1 +<br />
c=1<br />
ɛ c (<br />
c!<br />
ɛ n )<br />
π c Γ (n + 1)!<br />
2c,c(Ω) +<br />
π n − ɛn n!n<br />
π n vol(Ω).<br />
4.5 Another method to compute the measure of <strong>complex</strong> l<strong>in</strong>es<br />
meet<strong>in</strong>g a regular doma<strong>in</strong><br />
From Theorem 4.3.5 we can give an expression of the measure of <strong>complex</strong> l<strong>in</strong>es meet<strong>in</strong>g a<br />
regular doma<strong>in</strong> (just tak<strong>in</strong>g r = 1). Here, we give another method to obta<strong>in</strong> this expression,<br />
us<strong>in</strong>g the results <strong>in</strong> Chapter 3.<br />
4.5.1 Measure of <strong>complex</strong> l<strong>in</strong>es meet<strong>in</strong>g a regular doma<strong>in</strong> <strong>in</strong> C n<br />
Proposition 4.5.1. Let Ω ⊂ C n be a regular doma<strong>in</strong>. Then,<br />
∫<br />
χ(Ω ∩ L 1 )dL 1 =<br />
ω (<br />
∫ )<br />
2n−4<br />
(2n − 1)M 1 (∂Ω) + k n (Jn) .<br />
4n(n − 1)<br />
∂Ω<br />
L C 1<br />
Proof. Recall that each <strong>complex</strong> l<strong>in</strong>e is isometric to C. Gauss-Bonnet formula <strong>in</strong> C n for<br />
hypersurfaces ∂Ω states<br />
M 2n−1 (∂Ω) = 2nω 2n χ(Ω).<br />
Apply<strong>in</strong>g Gauss-Bonnet formula <strong>in</strong> C and Proposition 3.3.2 with s = 1 we get the result<br />
∫<br />
χ(Ω ∩ L 1 )dL 1 = 1 ∫ ∫<br />
k g dpdL 1 = ω (<br />
∫ )<br />
2n−2<br />
(2n − 1)M 1 (∂Ω) + k n (Jn) .<br />
L C 2π<br />
1<br />
L C 1 ∂Ω∩L 1<br />
4nω 2 ∂Ω
4.5 Another method to compute the measure of <strong>complex</strong> l<strong>in</strong>es 77<br />
Although Gauss-Bonnet formula is known <strong>in</strong> C n for n ≥ 1, we cannot apply the same<br />
method to give the expression of the measure of s-planes meet<strong>in</strong>g a regular doma<strong>in</strong> s<strong>in</strong>ce<br />
the <strong>in</strong>tegral ∫ L C r M 2r−1(∂Ω ∩ L r )dL r is not <strong>in</strong> general known. In the next section we get<br />
an expression for this <strong>in</strong>tegral us<strong>in</strong>g the Gauss-Bonnet formula and the measure of <strong>complex</strong><br />
r-planes meet<strong>in</strong>g a regular doma<strong>in</strong>.<br />
4.5.2 Measure of <strong>complex</strong> l<strong>in</strong>es meet<strong>in</strong>g a regular doma<strong>in</strong> <strong>in</strong> CP n and CH n<br />
The follow<strong>in</strong>g result is given, for <strong>in</strong>stance, <strong>in</strong> [ÁPF04].<br />
Proposition 4.5.2 ([ÁPF04]). Let Ω be a regular doma<strong>in</strong> <strong>in</strong> CPn or CH n . Then,<br />
∫<br />
vol 2s (Ω ∩ L s )dL s = Cvol 2n (Ω).<br />
L C s<br />
The value of the constant C, it is not known, but now we shall need it explicitly.<br />
Proposition 4.5.3. Let Ω be a regular doma<strong>in</strong> <strong>in</strong> CP n or CH n . Then,<br />
∫<br />
vol 2s (Ω ∩ L s )dL s = vol(G C n,n−s)vol 2n (Ω).<br />
L C s<br />
Proof. In order to f<strong>in</strong>d C we apply last proposition to a ball of radius R. Let L s be a <strong>complex</strong><br />
s-plane meet<strong>in</strong>g B R at a distance ρ from the center of the ball. From Lemma 3.2.13 <strong>in</strong> [Gol99],<br />
we have that the <strong>in</strong>tersection B R ∩ L s is a ball of <strong>complex</strong> dimension s and radius r such that<br />
cos ɛ (R) = cos ɛ (r) cos ɛ (ρ).<br />
The expression of the volume of a geodesic ball of radius R <strong>in</strong> CK n (ɛ) is (cf. [Gra73])<br />
vol 2n (B R ) =<br />
πn<br />
|ɛ| n n! s<strong>in</strong>2n ɛ (R).<br />
Us<strong>in</strong>g this expression we get<br />
vol 2s (L s ∩ B R ) =<br />
( ) π s<br />
1<br />
|ɛ| s!<br />
( cos<br />
2<br />
ɛ (R) − cos 2 ) s<br />
ɛ(ρ)<br />
cos 2 .<br />
ɛ(ρ)<br />
On the other hand, the Jacobian of the change of variables to polar coord<strong>in</strong>ates is given by<br />
(cf. [Gra73])<br />
cos ɛ (R) s<strong>in</strong>ɛ<br />
2n−1 (R)<br />
|ɛ| n−1/2 .<br />
Then, us<strong>in</strong>g Proposition 1.5.8, we get<br />
∫<br />
L C s<br />
vol 2s (L s ∩ B R )dL s = πs vol(G C n,n−s)O 2(n−s)−1<br />
|ɛ| n−1/2 s!<br />
·<br />
∫ R<br />
0<br />
s∑<br />
( ) s<br />
(−1) i+1 ·<br />
i<br />
i=0<br />
s<strong>in</strong> 2(n−s)−1<br />
ɛ (ρ) cos 2i<br />
ɛ (R) cos 2(s−i)+1<br />
ɛ (ρ)dρ<br />
= πs vol(G C n,n−s)O 2(n−s)−1<br />
|ɛ| n−1/2 ·<br />
s!<br />
s∑ ∑s−i<br />
i∑<br />
( )( )( 2(n−s+k+j)<br />
s s − i i s<strong>in</strong><br />
· (−1) i+1 ɛ (R)<br />
√ .<br />
i j k)<br />
ɛ(n − s + j)<br />
i=0 j=0 k=0
78 Gauss-Bonnet Theorem and Crofton formulas for <strong>complex</strong> planes<br />
From Proposition 4.5.2, this expression is a multiple of<br />
vol 2n (B R ) =<br />
πn<br />
|ɛ| n n! s<strong>in</strong>2n ɛ (R).<br />
Thus, all terms <strong>in</strong> the sum are zero except for 2(n − s + k + j) = 2n, i.e. k + j = s, which<br />
together with j ≤ s − i, k ≤ i implies j = s − i, k = i, and we get<br />
∫<br />
vol 2s (L s ∩ B R )dL s<br />
L C s<br />
F<strong>in</strong>ally, from equality<br />
= πs vol(G C n,n−s)O 2(n−s)−1<br />
|ɛ| n s!<br />
= πs vol(G C n,n−s)O 2(n−s)−1<br />
2|ɛ| n<br />
s∑<br />
( ) s s<strong>in</strong><br />
(−1) i+1 2n<br />
ɛ (R)<br />
i 2(n − i)<br />
i=0<br />
s<strong>in</strong> 2n<br />
ɛ (R)<br />
s∑<br />
i=0<br />
= πs vol(G C n,n−s)O 2(n−s)−1 (n − s − 1)!<br />
2|ɛ| n n!<br />
and us<strong>in</strong>g O 2(n−s)−1 = 2(n − s)ω 2(n−s) = 2<br />
(−1) i+1<br />
i!(s − i)!(n − i)<br />
s<strong>in</strong> 2n<br />
ɛ (R).<br />
C<br />
πn<br />
|ɛ| n n! = πs vol(G C n,n−s)O 2(n−s)−1 (n − s − 1)!<br />
2|ɛ| n ,<br />
n!<br />
πn−s<br />
(n−s−1)!<br />
, we get the value of the constant C.<br />
Corollary 4.5.4. Let Ω ⊂ CK n (ɛ) be a regular doma<strong>in</strong>. Then,<br />
∫<br />
χ(Ω ∩ L 1 )dL 1 =<br />
ω (<br />
∫<br />
)<br />
2n−4<br />
(2n − 1)M 1 (∂Ω)+ k n (JN)+8nɛvol(Ω) .<br />
4n(n − 1)<br />
∂Ω<br />
L C 1<br />
where k n (JN) denotes the normal curvature <strong>in</strong> the direction JN.<br />
Proof. Us<strong>in</strong>g Gauss-Bonnet formula <strong>in</strong> H 2 (−4) we have (cf. [San04, page 309])<br />
∫<br />
χ(Ω ∩ L 1 )dL 1 = 1 ∫<br />
M 1 (∂Ω ∩ L 1 )dL 1 − 2 ∫<br />
vol(Ω ∩ L 1 )dL 1 ,<br />
2π L C π<br />
1<br />
L C 1<br />
and us<strong>in</strong>g Proposition 4.5.3 with s = 1, and Proposition 3.3.2 we get the result.<br />
Corollary 4.5.5. If Ω is a regular doma<strong>in</strong> <strong>in</strong> CK 2 (ɛ), ɛ ≠ 0, then<br />
∫<br />
Ω∩L 1 ≠∅<br />
χ(∂Ω ∩ L 1 )dL 1 = 1 4<br />
(M 1 (∂Ω) − 1 3ɛ M 3(∂Ω) + 4ɛvol(Ω) + 2π2<br />
3ɛ χ(Ω) )<br />
.<br />
Proof. From previous corollary, with n = 2, we have<br />
∫<br />
χ(Ω ∩ L 1 )dL 1 = 1 (<br />
∫<br />
)<br />
3M 1 (∂Ω) + k n (JN) + 16ɛvol(Ω) .<br />
L C 8<br />
1 ∂Ω<br />
Isolat<strong>in</strong>g ∫ ∂Ω k n(JN) <strong>in</strong> expression (4.31) we get the stated result.<br />
Note that the previous corollary cannot be extended to ɛ = 0 s<strong>in</strong>ce the expression (4.31)<br />
does not conta<strong>in</strong> the term ∫ ∂Ω k n(JN).<br />
L C 1
4.6 Total Gauss curvature <strong>in</strong>tegral C n 79<br />
4.6 Total Gauss curvature <strong>in</strong>tegral C n<br />
Theorem 4.6.1. If Ω ⊂ C n is a regular doma<strong>in</strong>, then<br />
∫<br />
L C r<br />
( ) n − 1 −1 ( ) n<br />
M 2r−1 (∂Ω∩L r )dL r = 2rω2rvol(G 2 C n−1,r)<br />
−1·<br />
r r<br />
⎛<br />
⎞<br />
n−r<br />
∑<br />
( )<br />
· ⎝<br />
1 2n − 2r − 2q<br />
4 n−r−q µ 2n−2r,q (Ω) ⎠ .<br />
n − r − q<br />
q=max{0,n−2r}<br />
Proof. On one hand, by Gauss-Bonnet formula <strong>in</strong> C n and the relation (4.6), we have<br />
∫<br />
∫<br />
∫<br />
M 2r−1 (∂Ω ∩ L r )dL r = 2rω r χ(Ω ∩ L r )dL r = 2rω 2r µ 0,0 (Ω ∩ L r )dL r . (4.32)<br />
L C r<br />
On the other hand, by Theorem 4.3.1, we have<br />
∫<br />
L C r<br />
L C r<br />
( ) n − 1 −1 ( n −1<br />
χ(∂Ω ∩ L r )dL r = vol(G C n−1,r)ω 2r<br />
r r)<br />
⎛<br />
⎞<br />
n−r<br />
∑<br />
( )<br />
· ⎝<br />
1 2n − 2r − 2q<br />
4 n−r−q µ 2n−2r,q (Ω) ⎠ .<br />
n − r − q<br />
q=max{0,n−2r}<br />
If we equate both expressions and we use the relation (4.6), we get the result.<br />
L C r
Chapter 5<br />
Other Crofton formulas<br />
In the previous chapter we give an expression for the measure of <strong>complex</strong> planes <strong>in</strong>tersect<strong>in</strong>g<br />
a regular doma<strong>in</strong> <strong>in</strong> a <strong>complex</strong> <strong>space</strong> form. Complex planes <strong>in</strong> CK n (ɛ) are totally geodesic<br />
submanifolds, but, by Theorem 1.4.6, there are other totally geodesic submanifolds. Totally<br />
real planes are also totally geodesic submanifolds <strong>in</strong> CK n (ɛ) for any ɛ (cf. Theorem 1.4.6).<br />
Moreover, for ɛ = 0, all submanifolds generated by the exponential map of a vector sub<strong>space</strong><br />
holomorphically isometric to C k ⊕ R k−2p are totally geodesic. Note that <strong>complex</strong> planes and<br />
totally real planes are particular cases of these submanifolds, for (k, p) = (2p, p) and (k, p) =<br />
(k, 0), respectively.<br />
In this chapter we obta<strong>in</strong> an expression for the measure of planes of type (2n−p, n−p), the<br />
so-called coisotropic planes, <strong>in</strong>tersect<strong>in</strong>g a doma<strong>in</strong> <strong>in</strong> C n , and an expression for the measure of<br />
Lagrangian planes <strong>in</strong> CK n (ɛ).<br />
5.1 Space of (k, p)-planes<br />
First, we recall the def<strong>in</strong>ition of (k, p)-plane <strong>in</strong> C n , as it is given <strong>in</strong> [BF08].<br />
Def<strong>in</strong>ition 5.1.1. Suppose that V is a real vector <strong>space</strong> and L n k<br />
(V ) denote the <strong>space</strong> of all<br />
aff<strong>in</strong>e sub<strong>space</strong>s of dimension k <strong>in</strong> V . If V = C n , considered as a real vector <strong>space</strong>, then the<br />
<strong>space</strong> of (k, p)-planes, L k,p (C n ) ⊂ L n k (Cn ) is def<strong>in</strong>ed as the subset of all sub<strong>space</strong> of (real)<br />
dimension k that can be expressed as the orthogonal direct sum of a <strong>complex</strong> sub<strong>space</strong> of<br />
<strong>complex</strong> dimension p and a totally real sub<strong>space</strong> of (real) dimension (k − 2p).<br />
We denote the elements of L k,p (C n ) by L k,p and the Grassmannian of all (k, p)-planes<br />
through the orig<strong>in</strong> <strong>in</strong> a vector <strong>space</strong> V by G n,k,p (V ).<br />
From the previous def<strong>in</strong>ition, L k,p (C n ) is the orbit of C p ⊕ R k−2p under the action of<br />
C n ⋊ U(n) (which are the holomorphic isometries of C n ).<br />
The notion of (k, p)-plane was extended to CH n . In [Gol99] and [Hsi98], they are def<strong>in</strong>ed<br />
as particular cases of the so-called l<strong>in</strong>ear submanifolds.<br />
Def<strong>in</strong>ition 5.1.2. The image of the exponential map from a po<strong>in</strong>t x ∈ CH n of a vector<br />
sub<strong>space</strong> <strong>in</strong> T x CH n is called l<strong>in</strong>ear submanifold.<br />
The image of the exponential map from a po<strong>in</strong>t x ∈ CH n of a (k, p)-plane <strong>in</strong> T x CH n is<br />
called l<strong>in</strong>ear (k, p)-plane.<br />
This def<strong>in</strong>ition could be also stated <strong>in</strong> CP n , but do<strong>in</strong>g so, we obta<strong>in</strong> submanifolds with<br />
s<strong>in</strong>gularities except when the submanifold is totally geodesic, i.e. for <strong>complex</strong> planes (which<br />
correspond to (2p, p)-planes), and for totally real planes (which correspond to (k, 0)-planes).<br />
In CH n , l<strong>in</strong>ear submanifolds are not always totally geodesic submanifolds. They are totally<br />
geodesic just for <strong>complex</strong> planes and totally real planes (cf. Theorem 1.4.6).<br />
81
82 Other Crofton formulas<br />
5.1.1 Bisectors<br />
As <strong>in</strong> <strong>complex</strong> hyperbolic <strong>space</strong> there are no totally geodesic real hypersurfaces, it is natural<br />
to look for real hypersurfaces with similar properties to those expected for a totally geodesic<br />
real hypersurface. In [Gol99, page 152] it is answered that they are the so-called bisectors, also<br />
denoted by sp<strong>in</strong>al superfaces.<br />
Def<strong>in</strong>ition 5.1.3. Let z 1 , z 2 be two different po<strong>in</strong>ts <strong>in</strong> CH n . The bisector equidistant from z 1<br />
and z 2 is def<strong>in</strong>ed as<br />
E(z 1 , z 2 ) = {z ∈ CH n | d(z 1 , z) = d(z 2 , z)}<br />
where d(z, z i ) denotes the distance between po<strong>in</strong>ts z i and z at CK n (ɛ) (see Proposition 1.2.3).<br />
Def<strong>in</strong>ition 5.1.4. Let z 1 , z 2 be two different po<strong>in</strong>ts <strong>in</strong> CH n .<br />
• The <strong>complex</strong> geodesic Σ def<strong>in</strong>ed by z 1 and z 2 is the <strong>complex</strong> sp<strong>in</strong>e of the bisector E(z 1 , z 2 ).<br />
• The real sp<strong>in</strong>e of the bisector E(z 1 , z 2 ), σ(z 1 , z 2 ) is the <strong>in</strong>tersection between the bisector<br />
and the <strong>complex</strong> sp<strong>in</strong>e, i.e.<br />
σ(z 1 , z 2 ) = E(z 1 , z 2 ) ∩ Σ(z 1 , z 2 ) = {z ∈ Σ | d(z 1 , z) = d(z 2 , z)}.<br />
• A slide of E is a <strong>complex</strong> hyperplane Π −1<br />
Σ (s) where Π : CHn −→ Σ denotes the orthogonal<br />
projection over Σ.<br />
Remark 5.1.5. 1. The set of all slides <strong>in</strong> a bisector def<strong>in</strong>es a foliation of the bisector by<br />
<strong>complex</strong> hyperplanes.<br />
2. The real sp<strong>in</strong>e is a (real) geodesic <strong>in</strong> CH n s<strong>in</strong>ce Σ is totally geodesic and isometric to<br />
H 2 , and <strong>in</strong> the real hyperbolic <strong>space</strong>, the bisector l<strong>in</strong>e of two given po<strong>in</strong>ts is a geodesic.<br />
3. Each geodesic γ ⊂ CH n is the real sp<strong>in</strong>e of a unique bisector. Indeed, take the <strong>complex</strong><br />
l<strong>in</strong>e Σ conta<strong>in</strong><strong>in</strong>g γ and the orthogonal projection Π Σ to Σ. Then, Π −1<br />
Σ<br />
(γ) def<strong>in</strong>es a<br />
bisector.<br />
Example 5.1.6. In CH 2 with the projective model, the bisector with respect to z 1 = [(1, 0, i)]<br />
and z 2 = [(1, 0, −i)] is<br />
E(z 1 , z 2 ) = {[(1, z, t)] ∈ CH n | z ∈ C, t ∈ R}.<br />
This expression can be obta<strong>in</strong>ed directly us<strong>in</strong>g the formula for the distance between 2 po<strong>in</strong>ts<br />
given at Proposition 1.2.3. {[(<br />
The <strong>complex</strong> sp<strong>in</strong>e is 1, 0, i λ − µ )]<br />
}<br />
with λ, µ ∈ C both nonzero and the real sp<strong>in</strong>e<br />
λ + µ<br />
is {[1, 0, t]}.<br />
Proposition 5.1.7. The isometries of CH n act transitively over the <strong>space</strong> of bisectors.<br />
Proof. Us<strong>in</strong>g the correspondence between bisectors and real geodesics, we have that isometries<br />
act transitively over the <strong>space</strong> of bisectors, s<strong>in</strong>ce they do so over the <strong>space</strong> of real geodesics.<br />
It is known that there are no non-trivial isometries which fix po<strong>in</strong>twise a bisector , s<strong>in</strong>ce<br />
they are not totally geodesic hypersurfaces (if ɛ ≠ 0). Anyway, we can consider the reflection<br />
with respect to a slice S of the bisector. This reflexion fixes po<strong>in</strong>twise the slice S and lies the<br />
bisector <strong>in</strong>variant. Moreover, each of these reflexions is also a reflection with respect to the<br />
sp<strong>in</strong>e σ, thus, it fixes the po<strong>in</strong>ts <strong>in</strong> σ ∩ S.
5.1 Space of (k, p)-planes 83<br />
Bisectors are hypersuperfaces of cohomogenity one. That is, the orbit, for the group of the<br />
isometries fix<strong>in</strong>g a bisector, of a po<strong>in</strong>t (except for the po<strong>in</strong>ts <strong>in</strong> the real sp<strong>in</strong>e, which are of null<br />
measure <strong>in</strong> the bisector) <strong>in</strong> a bisector is a submanifold of codimension 1 <strong>in</strong>side the bisector; a<br />
submanifold of codimension 2 <strong>in</strong> CH n (cf. [GG00]).<br />
Thus, bisectors are not homogeneous hypersurfaces. By def<strong>in</strong>ition a hypersurface is homogeneous<br />
if the orbit of each po<strong>in</strong>t (for the isometry group fix<strong>in</strong>g the hypersurface) is all the<br />
hypersurface.<br />
Let us study the orbit of a po<strong>in</strong>t <strong>in</strong> a bisector.<br />
A geodesic γ(t) ⊂ CH n uniquely determ<strong>in</strong>es a tube T (r) of radius r and a bisector E with<br />
real sp<strong>in</strong>e γ.<br />
We study the relation between these two hypersurfaces, obta<strong>in</strong><strong>in</strong>g the orbit of a po<strong>in</strong>t <strong>in</strong><br />
the bisector <strong>in</strong> terms of the tube conta<strong>in</strong><strong>in</strong>g the po<strong>in</strong>t.<br />
Proposition 5.1.8. Let p ∈ E. Consider r such that p ∈ T (r) ∩ E where T (r) denotes the<br />
tube of radius r along the real sp<strong>in</strong>e γ of E. The orbit of p by the isometries fix<strong>in</strong>g E, is given<br />
by T (r) ∩ E.<br />
Proof. Each po<strong>in</strong>t of the orbit O p of p belongs to T (r) s<strong>in</strong>ce the isometries fix<strong>in</strong>g the bisector<br />
fix the sp<strong>in</strong>e, and they preserve distances. Then, O p ⊂ T (r) ∩ E.<br />
Every po<strong>in</strong>t <strong>in</strong> T (r) ∩ E belongs to the orbit of p. Indeed, if q ∈ T (r) ∩ E then d(q, γ(t)) =<br />
d(p, γ(t)), a necessary condition to be q <strong>in</strong> the orbit of p. The projection of the po<strong>in</strong>ts p, q to<br />
Σ can or cannot be the same po<strong>in</strong>t. Let us prove that <strong>in</strong> both cases there exists an isometry<br />
g such that fixes the bisector and g(p) = q.<br />
Suppose that p and q project at Σ to the same po<strong>in</strong>t x. Then p and q belong to the<br />
same slide of E. Let us def<strong>in</strong>e an isometry g fix<strong>in</strong>g x and the bisector. We denote by v the<br />
tangent vector to the real sp<strong>in</strong>e γ at x. As the isometries fix<strong>in</strong>g the bisector also fix γ, g<br />
satisfies dg(v) = ±v. Moreover, as isometries preserve the holomorphic angle dg(Jv) = ±Jv.<br />
Thus, g fixes the <strong>complex</strong> sp<strong>in</strong>e Σ and its orthogonal complement at x, which is the slide<br />
conta<strong>in</strong><strong>in</strong>g p and q, and is isometric to CH n−1 . Now, <strong>in</strong> CH n−1 there exists an isometry ˜g<br />
such that ˜g(p) = q (s<strong>in</strong>ce CH n−1 is a homogeneous <strong>space</strong>). Therefore, g def<strong>in</strong>ed by g(x) = x,<br />
dg(v) = ±v, dg(Jv) = ±Jv and dg(u) = ˜dg(u), for all u ∈ 〈v, Jv〉 ⊥ , gives an isometry of<br />
CK n (ɛ) fix<strong>in</strong>g E and such that g(p) = q.<br />
Suppose that p and q do not project at Σ to the same po<strong>in</strong>t. Let x = Π Σ p and y = Π Σ q.<br />
Note that x, y ∈ γ s<strong>in</strong>ce p and q are po<strong>in</strong>ts <strong>in</strong> the bisector. Then, there exists a reflection ρ<br />
such that ρ(x) = y and ρ(γ) = γ. Thus, dρ takes the orthogonal <strong>space</strong> of {γ x, ′ Jγ x} ′ to the<br />
orthogonal <strong>space</strong> of {γ y, ′ Jγ y}. ′ Moreover, ˜q = ρ(q) satisfies Π Σ˜q = Π Σ q. If we consider the<br />
po<strong>in</strong>ts ˜q and q, then we are <strong>in</strong> the previous case and we know that there exists an isometry g<br />
such that g(˜q) = q.<br />
From this proposition we have that the subset of bisectors conta<strong>in</strong><strong>in</strong>g a po<strong>in</strong>t is a noncompact<br />
set, <strong>in</strong> the <strong>space</strong> of bisectors. In the next proposition, we prove that the measure of<br />
bisectors meet<strong>in</strong>g a regular doma<strong>in</strong> is <strong>in</strong>f<strong>in</strong>ite.<br />
Remark 5.1.9. Denote by dL the <strong>in</strong>variant density of the <strong>space</strong> of bisectors B and by dL 1<br />
the <strong>in</strong>variant density of the <strong>space</strong> of real geodesics <strong>in</strong> CH n . By the correspondence between<br />
geodesics and bisectors we have<br />
dL = dL 1 .<br />
If Σ denotes the <strong>complex</strong> l<strong>in</strong>e conta<strong>in</strong><strong>in</strong>g a real geodesic γ, then the density of the <strong>space</strong><br />
of real geodesics can be expressed as<br />
dL 1 = dL Σ 1 dΣ,
84 Other Crofton formulas<br />
where dΣ denotes the <strong>in</strong>variant density of <strong>complex</strong> l<strong>in</strong>es and dL Σ 1 the <strong>in</strong>variant density of the<br />
<strong>space</strong> of real geodesics conta<strong>in</strong>ed <strong>in</strong> Σ. Us<strong>in</strong>g polar coord<strong>in</strong>ates ρ, θ <strong>in</strong> Σ we get<br />
dL Σ 1 dΣ = cos 4ɛ (ρ)dρdθdΣ. (5.1)<br />
Proposition 5.1.10. The measure of bisectors meet<strong>in</strong>g a regular doma<strong>in</strong> <strong>in</strong> CK n (ɛ) with ɛ < 0<br />
is <strong>in</strong>f<strong>in</strong>ite.<br />
Proof. We prove that the measure of bisectors <strong>in</strong>tersect<strong>in</strong>g a ball B of radius R <strong>in</strong> CK n (ɛ) is<br />
<strong>in</strong>f<strong>in</strong>ite, i.e.<br />
∫<br />
χ(B ∩ L)dL = +∞.<br />
B<br />
Consider the expression for the density of bisectors <strong>in</strong> (5.1). Denote the volume element<br />
of CK n (ɛ) by dx. Fixed a bisector L by x, then dx can be expressed as dx = dx 1 dx L where<br />
dx L denotes the volume element <strong>in</strong> the bisector and dx 1 the length element <strong>in</strong> the direction<br />
N x orthogonal to the bisector at x.<br />
If N y is the normal vector to the bisector at y = Π Σ (x), then the plane spanned by N y<br />
(which co<strong>in</strong>cides with the normal vector to the real sp<strong>in</strong>e <strong>in</strong>side Σ) and the tangent vector u<br />
to the geodesic jo<strong>in</strong><strong>in</strong>g y and x is a totally real plane and conta<strong>in</strong>s N x .<br />
Thus, the plane exp y (span{N y , u}) is isometric to H 2 (ɛ). If r denotes the distance between<br />
y and x, then dx 1 = cos ɛ (r)dy 1 where dy 1 denotes the length element <strong>in</strong> the direction N y .<br />
In the previous remark, we give an expression for dL 1 . Now, we use it tak<strong>in</strong>g polar<br />
coord<strong>in</strong>ates with respect to y ∈ Σ, so ρ = 0. Then, dy 1 = dρ and<br />
dx L dL 1 = dx L dL Σ 1 dΣ = dx L dθdρdΣ = dθdx L dy 1 dΣ =<br />
1<br />
cos ɛ (r) dθdΣdx.<br />
On the other hand, fixed a regular doma<strong>in</strong> Ω ⊂ CK n (ɛ) it follows, for some constant C > 0,<br />
vol(Ω) < 1 C χ(Ω).<br />
Then,<br />
∫<br />
∫<br />
∫ ∫<br />
χ(B ∩ L)dL > C vol(B ∩ L)dL = C dx L dL<br />
B<br />
∫<br />
= C<br />
B<br />
B<br />
∫<br />
(1.16)<br />
= 2πC<br />
L C 1<br />
∫<br />
∫ 2π<br />
B<br />
∫<br />
= 2πvol(B)<br />
0<br />
∫<br />
B<br />
B∩L<br />
∫<br />
1<br />
cos ɛ (r)<br />
∫B<br />
dθdΣdx = 2πC ∫<br />
G C n,n−1<br />
CH n−1 (ɛ)<br />
L C (n−1)[x]<br />
L C 1<br />
1<br />
cos ɛ (r) dΣdx<br />
cos 2 ɛ(r)<br />
cos ɛ (r) dp n−1dG n,n−1 dx<br />
cos ɛ (r)dx = +∞.<br />
5.2 Variation of the measure of planes meet<strong>in</strong>g a regular doma<strong>in</strong><br />
At Chapter 4 we give an expression for the measure of <strong>complex</strong> r-planes meet<strong>in</strong>g a regular<br />
doma<strong>in</strong> <strong>in</strong> CK n (ɛ). Now, we give a generalization of this result for the <strong>space</strong> of (k, p)-planes<br />
<strong>in</strong> C n , and for the <strong>space</strong> of totally real k-planes <strong>in</strong> CK n (ɛ).<br />
First, we need the expression of the density of the <strong>space</strong> of (k, p)-planes with respect to<br />
the <strong>forms</strong> ω ij def<strong>in</strong>ed at (1.13).
5.2 Variation of the measure of planes meet<strong>in</strong>g a regular doma<strong>in</strong> 85<br />
Lemma 5.2.1.<br />
1. In C n , the <strong>space</strong> L k,p is a homogeneous <strong>space</strong> and<br />
L k,p<br />
∼ = U(n) ⋉ C n /(U(p) × O(k − 2p) × U(n − k + p) ⋉ R n ).<br />
Let {g; g 1 , g 2 , ..., g 2n−1 , g 2n } with g 2i = Jg 2i+1 be a J-mov<strong>in</strong>g frame adapted to a (k, p)-<br />
plane <strong>in</strong> g such that {g 1 , Jg 1 , ..., g p , Jg p , g 2p+1 , ..., g 2k−1 } expand the tangent <strong>space</strong> to the<br />
(k, p)-plane. The <strong>in</strong>variant density of L k,p is given by<br />
∣ ∣∣∣∣∣ ∧ ∧<br />
dL k,p =<br />
ω i ω ji (5.2)<br />
∣<br />
i<br />
where i ∈ {2p + 2, 2p + 4, ..., 2k, 2k + 1, 2k + 2, ..., 2n} and j ∈ {1, 3, ..., 2p − 1, 2p + 1, 2p +<br />
3, ..., 2k − 1}.<br />
2. In CK n (ɛ), ɛ ≠ 0, the <strong>space</strong> of <strong>complex</strong> p-planes L C p and the <strong>space</strong> of totally real k-planes<br />
L R k<br />
are homogeneous <strong>space</strong>s and<br />
j,i<br />
L C p ∼ = U ɛ (n)/(U ɛ (p) × U(n − p)),<br />
where<br />
L R k ∼ = U ɛ (n)/(O ɛ (k) × U(n − k)),<br />
U ɛ (n) =<br />
{ U(1 + n), if ɛ > 0,<br />
U(1, n), if ɛ < 0.<br />
, O ɛ (k) =<br />
{ O(1 + k), if ɛ > 0,<br />
O(1, k), if ɛ < 0.<br />
Moreover, fixed a J-mov<strong>in</strong>g frame as <strong>in</strong> the previous statement, the expression (5.2)<br />
rema<strong>in</strong>s true.<br />
Proof. 1. By Lemma 1.5.1 we have that the isometry group of C n acts transitively over<br />
J-basis. Thus, there exists an isometry that carries a fixed (k, p)-plane to another.<br />
The isotropy group of a (k, p)-plane <strong>in</strong> C n is isomorphic to U(p)×O(k−2p)×U(n−k+p)<br />
s<strong>in</strong>ce (k, p)-planes <strong>in</strong> C n are totally geodesic submanifolds and the tangent <strong>space</strong> at each<br />
po<strong>in</strong>t is isometric to C p ⊕ R k−2p .<br />
The density can be obta<strong>in</strong>ed us<strong>in</strong>g the theory of mov<strong>in</strong>g frames that we have discussed<br />
<strong>in</strong> Section 1.3.<br />
2. The arguments for the previous case are also valid, s<strong>in</strong>ce we restrict to totally geodesic<br />
submanifolds.<br />
The follow<strong>in</strong>g proposition is a generalization of the Proposition 4.2.1 for any (k, p)-plane<br />
<strong>in</strong> C n .<br />
Proposition 5.2.2. Let Ω ⊂ C n be a regular doma<strong>in</strong>, X a smooth vector field def<strong>in</strong>ed at C n<br />
with φ t the flow associated to X and Ω t = φ t (Ω). Then, <strong>in</strong> the <strong>space</strong> of (k, p)-planes L k,p <strong>in</strong><br />
C n , it is satisfied<br />
∫<br />
∫<br />
d<br />
dt∣ χ(Ω t ∩ L k,p )dL k,p = 〈∂φ/∂t, N〉<br />
σ k (II| V )dV dx<br />
t=0<br />
∫L k,p<br />
∂Ω<br />
G n,k,p (T x∂Ω)<br />
where N is the outward normal field at ∂Ω and σ k (II| V ) denotes the k-th symmetric elementary<br />
function of II restricted to V ∈ G n,k,p (T x ∂Ω), the Grassmanian of the (k, p)-planes conta<strong>in</strong>ed<br />
<strong>in</strong> the tangent <strong>space</strong> of ∂Ω at x.
86 Other Crofton formulas<br />
It is also satisfied the follow<strong>in</strong>g extension of Proposition 4.2.1 for totally real planes <strong>in</strong><br />
CK n (ɛ).<br />
Proposition 5.2.3. Let Ω ⊂ CK n (ɛ) be a regular doma<strong>in</strong>, X a smooth vector field def<strong>in</strong>ed at<br />
CK n (ɛ) with φ t the flow associated to X and Ω t = φ t (Ω). Then, <strong>in</strong> the <strong>space</strong> of totally real<br />
k-plane L R k <strong>in</strong> CKn (ɛ), it is satisfied<br />
∫<br />
d<br />
dt∣ t=0<br />
L R k<br />
∫<br />
∫<br />
χ(Ω t ∩ L k,p )dL k,p = 〈∂φ/∂t, N〉<br />
σ k (II| V )dV dx<br />
∂Ω<br />
G n,k,0 (T x∂Ω)<br />
where N is the outward normal field at ∂Ω and σ k (II| V ) denotes the k-th symmetric elementary<br />
function of II restricted to V ∈ G n,k,0 (T x ∂Ω), the Grassmanian of the totally real k-planes<br />
conta<strong>in</strong>ed <strong>in</strong> the tangent <strong>space</strong> of ∂Ω at x.<br />
Proof. This proof is analog to the proof of Proposition 4.2.1, s<strong>in</strong>ce the expression for the density<br />
of the <strong>space</strong> of totally real planes <strong>in</strong> (5.2) holds. We just have to modify the construction of<br />
the map γ <strong>in</strong> (4.7).<br />
For every x ∈ ∂Ω consider the curve c(t) = ϕ t (x). For every t, let D c(t) = 〈N c(t) , JN c(t) 〉 ⊥ ⊂<br />
dϕ t (T x ∂Ω) the <strong>complex</strong> hyperplane tangent to ϕ t (∂Ω) at c(t). If ∇ ∂t denotes the covariant<br />
derivative of CK n (ɛ) along c(t), we def<strong>in</strong>e<br />
∇ D ∂t X(t) = π t(∇ ∂t X(t))<br />
where π t : T c(t) CK n (ɛ) → D c(t) denotes the orthogonal projection. Given a vector X ∈ T x ∂Ω,<br />
there exists a unique vector field X(t) def<strong>in</strong>ed along c(t) such that ∇ D ∂tX(t) = 0 (it can be<br />
proved <strong>in</strong> the same way as the existence of the usual parallel translation). This def<strong>in</strong>e a l<strong>in</strong>ear<br />
map ψ t : D x → D c(t) , which preserves the <strong>complex</strong> structure J s<strong>in</strong>ce<br />
∇ D ∂t JX(t) = π t(∇ ∂t JX(t)) = π t (J∇ ∂t X(t)) = Jπ t (∇ ∂t X(t)).<br />
F<strong>in</strong>ally, we extend ψ t l<strong>in</strong>early to ψ t : T x ∂Ω → dϕ t (T x ∂Ω) such that ψ t (JN x ) = JN c(t) . This<br />
map takes totally real planes <strong>in</strong>to totally real planes. So, we can def<strong>in</strong>e the new map γ as<br />
γ : G n,k,p (T ∂Ω) × (−ɛ, ɛ) −→ L k,p<br />
((x, V ), t) ↦→ exp φt(x) ψ t (V ) .<br />
5.3 Measure of real geodesics <strong>in</strong> CK n (ɛ)<br />
The follow<strong>in</strong>g result, obta<strong>in</strong>ed straightforward from the last proposition, states that <strong>in</strong> <strong>complex</strong><br />
<strong>space</strong> <strong>forms</strong>, the measure of real geodesics meet<strong>in</strong>g a regular doma<strong>in</strong> is a multiple of the area<br />
of the doma<strong>in</strong> (as <strong>in</strong> real <strong>space</strong> <strong>forms</strong>).<br />
Theorem 5.3.1. Let Ω ⊂ CK n (ɛ) be a regular doma<strong>in</strong>, let X be a smooth vector field over<br />
CK n (ɛ), let φ t be the flow associated to X and let Ω t = φ t (Ω). Then<br />
∫<br />
d<br />
dt∣ χ(Ω t ∩ L 1 )dL R 1 = O 2n+1 ( ˜B 2n−2,n−2 (Ω) + ˜Γ 2n−2,n−1 (Ω))<br />
t=0<br />
L R 1<br />
and<br />
∫<br />
L R 1<br />
χ(Ω ∩ L 1 )dL R 1 = ω 2n µ 2n−1,n−1 (Ω) = ω 2n<br />
2 vol(∂Ω).
5.4 Measure of real hyperplanes <strong>in</strong> C n 87<br />
Proof. From Proposition 5.2.2 we have<br />
∫<br />
∫<br />
d<br />
dt∣ χ(Ω t ∩ L 1 )dL R 1 =<br />
t=0<br />
L R 1<br />
∂Ω<br />
∫<br />
〈∂φ/∂t, N〉<br />
σ 1 (II| V )dV dx.<br />
G R n,1,0 (Tx∂Ω)<br />
Let us study the <strong>in</strong>tegral with respect to the Grassmanian of geodesics <strong>in</strong> the tangent <strong>space</strong><br />
of each po<strong>in</strong>t x ∈ ∂Ω. We denote by {f 1 , . . . , f 2n−1 } the pr<strong>in</strong>cipal directions at x. Then,<br />
∫<br />
σ 1 (II| V )dV = 1 ∫<br />
k n (v)dv = 1 2n−1<br />
∑<br />
∫<br />
〈v, f i 〉 2 k i dv<br />
G R n,1,0 (Tx∂Ω) 2 S 2n−1 2<br />
i=1<br />
S 2n−1<br />
= 1 2n−1<br />
∑<br />
∫<br />
k i 〈v, f i 〉 2 dv (3.1.2)<br />
= O 2n−1<br />
∑<br />
2n−1<br />
k i = O 2n−1(2n − 1)<br />
tr(II).<br />
2<br />
S 2n−1 4n<br />
4n<br />
i=1<br />
Thus, by Examples 2.4.20.3 and 2.4.20.4<br />
∫<br />
d<br />
dt∣ χ(Ω t ∩ L 1 )dL R 1 = O ∫<br />
2n−1(2n + 1)<br />
〈X, N〉tr(II)dx<br />
t=0 L R 4n<br />
1<br />
∂Ω<br />
= O ∫<br />
(<br />
2n−1<br />
〈X, N〉<br />
4n<br />
= O 2n−1<br />
4 n!<br />
=<br />
ω 2n<br />
2 (n − 1)!<br />
N(Ω)<br />
i=1<br />
1<br />
(n − 1)! γ ∧ θn−1 2 +<br />
( ∫<br />
〈X, N〉γ ∧ θ2 n−1 + (n − 1)<br />
N(Ω)<br />
N(Ω)<br />
(˜Γ′ 2n−2,n−1 (Ω) + (n − 1) ˜B 2n−2,n−2(Ω))<br />
′<br />
= O 2n+1 ( ˜B 2n−2,n−2 (Ω) + ˜Γ 2n−2,n−1 (Ω)).<br />
)<br />
1<br />
(n − 2)! β ∧ θ 1 ∧ θ2<br />
n−2<br />
∫<br />
)<br />
〈X, N〉β ∧ θ 1 ∧ θ n−2<br />
In C n , the valuation ∫ L<br />
χ(∂Ω∩L R 1 )dL 1 has degree 2n−1, so, it is a multiple of B 2n−1,n−1 (Ω),<br />
1<br />
which has variation (cf. Proposition 4.1.7)<br />
δ X µ 2n−1,n−1 (Ω) = c n,2n−1,n−1 (2c −1<br />
n,2n−2,n−1˜Γ 2n−2,n−1 (Ω) + c −1<br />
n,2n−2,n−2 (n − 1) ˜B 2n−2,n−2 (Ω))<br />
= c n,2n−2,n−1 (˜Γ ′ 2n−2,n−1(Ω) + (n − 1) ˜B 2n−2,n−2(Ω))<br />
′<br />
1<br />
= (˜Γ ′<br />
(n − 1)!ω<br />
2n−2,n−1(Ω) + (n − 1) ˜B 2n−2,n−2(Ω)).<br />
′<br />
1<br />
Therefore, compar<strong>in</strong>g both variations we get the stated result <strong>in</strong> C n . The same expression<br />
holds for ɛ ≠ 0 s<strong>in</strong>ce the variation of µ 2n−1,n−1 does not depend on ɛ.<br />
The relation with the volume of the boundary of the doma<strong>in</strong> is obta<strong>in</strong>ed from the relation<br />
of β 2n−1,n−1 with the second fundamental form given <strong>in</strong> Example 2.4.20.5.<br />
5.4 Measure of real hyperplanes <strong>in</strong> C n<br />
The measure of real hyperplanes <strong>in</strong>tersect<strong>in</strong>g a regular doma<strong>in</strong> <strong>in</strong> C n also follows immediately<br />
from Proposition 5.2.2. This particular case has <strong>in</strong>terest by its own s<strong>in</strong>ce real hyperplanes are<br />
submanifolds of codimension 1.<br />
Theorem 5.4.1. Let Ω ⊂ C n be a regular doma<strong>in</strong>, X a smooth vector field over C n , φ t the<br />
flow associated to X and Ω t = φ t (Ω). Then<br />
d<br />
dt∣ χ(Ω t ∩ L 2n−1,n−1 )dL 2n−1,n−1 = O 2n+1˜Γ0,0 (Ω)<br />
t=0<br />
∫L 2n−1,n−1<br />
and<br />
∫<br />
L 2n−1,n−1<br />
χ(Ω ∩ L 2n−1,n−1 )dL 2n−1,n−1 = ω 2n−1 µ 1,0 (Ω).<br />
2
88 Other Crofton formulas<br />
Proof. From Proposition 4.2.1 we have<br />
∫ ∫<br />
d<br />
dt∣ χ(Ω t ∩ L 2n−1,n−1 )dL 2n−1,n−1 = 〈X, N〉<br />
σ 2n−1 (II| V )dV dx<br />
t=0<br />
∫L 2n−1,n−1 ∂Ω 0 G n,2n−1,n−1 (T<br />
∫<br />
x∂Ω)<br />
= 〈X, N〉σ 2n−1 (II)dx<br />
∫∂Ω<br />
1<br />
= 〈X, N〉<br />
(n − 1)! γ ∧ θn−1 0<br />
N(Ω)<br />
= 2c−1 n,0,0,<br />
(n − 1)! ˜Γ 0,0 (Ω).<br />
In C n , the valuation ∫ L 2n−1,n−1<br />
χ(∂Ω ∩ L 2n−1,n−1 )dL 2n−1,n−1 has degree 1, thus, it is a<br />
multiple of µ 1,0 , which has variation<br />
δ X µ 1,0 = 2c n,1,0 c −1<br />
n,0,0˜Γ 0,0 .<br />
Then, compar<strong>in</strong>g both expressions we obta<strong>in</strong> the result.<br />
Remark 5.4.2. From Example 2.4.20, it follows the equality<br />
µ 1,0 (Ω) = 1 det(II| D ) =<br />
ω 2n−1<br />
∫∂Ω<br />
1 M<br />
ω<br />
2n−2(∂Ω).<br />
D<br />
2n−1<br />
On the other hand, there is just one l<strong>in</strong>early <strong>in</strong>dependent valuation <strong>in</strong> the <strong>space</strong> of cont<strong>in</strong>uous<br />
translation and U(n)-<strong>in</strong>variant valuations <strong>in</strong> C n of degree 1. Thus,<br />
µ 1,0 (Ω) = cM 2n−2 (∂Ω),<br />
and the measure of real hyperplanes <strong>in</strong> C n meet<strong>in</strong>g a regular doma<strong>in</strong> is a multiple of the<br />
so-called “mean width”, as <strong>in</strong> the Euclidean <strong>space</strong>.<br />
5.5 Measure of coisotropic planes <strong>in</strong> C n<br />
A sub<strong>space</strong>s of C n is called coisotropic if its orthogonal is a totally real plane.<br />
Lemma 5.5.1. The (2n − p, n − p)-planes <strong>in</strong> C n are the coisotropic planes.<br />
Proof. If L ∈ L 2n−p,n−p , then L ⊥ has dimensió 2n − (2n − p) = p. The dimension of the<br />
maximal <strong>complex</strong> sub<strong>space</strong> conta<strong>in</strong>ed <strong>in</strong> L ⊥ is n − (n − p) − p = 0. Thus, L ⊥ is a totally real<br />
plane.<br />
Reciprocally, if L ⊥ is a totally real p-plane, then L has dimension 2n − p and the maximal<br />
<strong>complex</strong> sub<strong>space</strong> has dimension n − p.<br />
Lemma 5.5.2. Let S ⊂ C n be a hypersurface and L ∈ L 2n−p,n−p , p ∈ {1, . . . , n}, be a<br />
(2n−p, n−p)-plane tangent <strong>in</strong> S at p. If N denotes a normal vector to S at x, then JN ∈ T x L.<br />
Proof. As L is a (2n−p, n−p)-plane, we can consider, at each po<strong>in</strong>t, a basis of its tangent <strong>space</strong><br />
of the form {e 1 , Je 1 , . . . , e n−p , Je n−p , e n−p+1 , e n−p+2 , . . . , e n }, <strong>in</strong> a way such that Je i ⊥T x L for<br />
i ∈ {n−p+1, . . . , n}. Moreover, we can complete this basis to a basis of T x C n with the vectors<br />
{J en−p+1 , . . . , Je n }.<br />
On the other hand, at x ∈ L ∩ S, is it satisfied T x L ⊂ T x S, thus, N⊥T x L, i.e.<br />
〈N, e i 〉 = 0, ∀ i ∈ {1, . . . , n}, (5.3)<br />
〈N, Je j 〉 = 0,<br />
∀ j ∈ {1, . . . , n − p}.<br />
Now, if JN = ∑ n<br />
i=1 α ie i + ∑ n<br />
i=1 β iJe i , then N = − ∑ n<br />
i=1 α iJe i + ∑ n<br />
i=1 β ie i . Us<strong>in</strong>g (5.3)<br />
we get JN = ∑ n<br />
i=n−p+1 α ie i . Thus, JN ∈ T x L.
5.5 Measure of coisotropic planes <strong>in</strong> C n 89<br />
From this lemma, we can prove the follow<strong>in</strong>g result.<br />
Theorem 5.5.3. Let Ω ⊂ C n be a regular doma<strong>in</strong>, X a smooth vector field def<strong>in</strong>ed <strong>in</strong> C n , φ t<br />
the flow associated to X and Ω t = φ t (Ω). Then,<br />
∫<br />
d<br />
dt∣ χ(Ω t ∩ L)dL = vol(G ( )<br />
n,2n−p,n−p)ω 2n−p+1<br />
n<br />
(2n − p + 2)<br />
−1· (5.4)<br />
t=0 L 2n−p,n−p<br />
(n − 1)!<br />
p − 1<br />
and<br />
∫<br />
·<br />
⌊ p−1<br />
2 ⌋ ∑<br />
q=max{0,p−n−1}<br />
4 q−p+1 ( ) 2n + 2q − 2p + 1 −1˜Γp−1,q (Ω),<br />
(2n + 2q − 2p + 3) n + q − p + 1<br />
χ(Ω ∩ L)dL = vol(G n,2n−p,n−p)ω 2n−p<br />
L 2n−p,n−p<br />
(n − 1)!<br />
·<br />
⌊ p 2 ⌋ ∑<br />
q=max{0,p−n}<br />
( ) n −1·<br />
p − 1<br />
( ) 2n + 2q − 2p − 1 −1<br />
4 q−p<br />
n + q − p − 1 2n + 2q − 2p + 1 µ p,q(Ω).<br />
Proof. First of all, we prove that for the <strong>space</strong> of coisotropics planes, the variation of the<br />
measure does not have contribution <strong>in</strong> ˜B p,q . From Proposition 5.2.2 we have<br />
∫<br />
∫<br />
d<br />
dt∣ χ(Ω t ∩ L)dL = 〈∂φ/∂t, N〉<br />
σ 2n−p (II| V )dV dx<br />
t=0<br />
∫L 2n−p,n−p<br />
∂Ω<br />
G n,2n−p,n−p (T x∂Ω)<br />
but each V ∈ G n,2n−p,n−p (T x ∂Ω), by the previous lemma, conta<strong>in</strong>s the JN direction (with N<br />
the outward normal vector to ∂Ω at x), so that II| V always conta<strong>in</strong>s the entry correspond<strong>in</strong>g<br />
to the normal curvature of the direction JN. From Lemma 2.4.18 we have that only the<br />
polynomials obta<strong>in</strong>ed from φ ∗ (γ k,q ) conta<strong>in</strong> this entry of the second fundamental form.<br />
∫<br />
In order to f<strong>in</strong>d the constants, we solve a l<strong>in</strong>ear system. First, note that the functional<br />
L 2n−p,n−p<br />
χ(Ω ∩ L)dL is a valuation <strong>in</strong> C n with homogeneous degree p. Thus, it can be<br />
expressed as a l<strong>in</strong>ear comb<strong>in</strong>ation of the Hermitian <strong>in</strong>tr<strong>in</strong>sic volumes with the same degree<br />
∫<br />
L 2n−p,n−p<br />
χ(Ω ∩ L)dL =<br />
⌊ p−1<br />
2 ⌋ ∑<br />
q=max{0,p−n}<br />
A p,q µ p,q (Ω) (5.5)<br />
for some A p,q , which we want to determ<strong>in</strong>e.<br />
Tak<strong>in</strong>g the variation <strong>in</strong> both sides, we f<strong>in</strong>d the value of these constants. By Proposition<br />
4.1.7, the variation on the right hand side of (5.5) is<br />
⌊ p 2 ⌋−1 ∑<br />
q=max{0,p−n−1}<br />
(A p,q 2c n,p,q c −1<br />
n,p−1,q (p − 2q)2 (5.6)<br />
− A p,q+1 2c n,p,q+1 c −1<br />
n,p−1,q (n − p + q + 1)(q + 1))˜Γ p−1,q<br />
+ (A p,q+1 2c n,p,q+1 c −1<br />
n,p−1,q (n − p + q + 3/2)(q + 1)<br />
− A p,q 2c n,p,q c −1<br />
n,p−1,q (p − 2q)(p − 2q − 1)) ˜B p−1,q<br />
+ A p,⌊<br />
p<br />
2 ⌋ 2c n,p,⌊<br />
p<br />
2 ⌋ c n,p−1,⌊<br />
p<br />
2 ⌋ (p − 2⌊ p 2 ⌋)2˜Γp−1,⌊ p<br />
2 ⌋ .<br />
Impos<strong>in</strong>g that the variation vanishes on ˜B p−1,q we get some equations, from which we obta<strong>in</strong><br />
the relations<br />
A p,q+1 = n − p + q + 1<br />
(n − p + q + 3/2) A p,q. (5.7)
90 Other Crofton formulas<br />
So, each A p,q+1 is a multiple of A p,max{0,p−n} . To f<strong>in</strong>d this value, we need another equation,<br />
obta<strong>in</strong>ed tak<strong>in</strong>g II| D = λId, λ ∈ R + , and equation the expression (5.6) with the variation of<br />
the Proposition 5.2.2. Then, for each (n, r) we have a compatible l<strong>in</strong>ear system, s<strong>in</strong>ce constant<br />
<strong>in</strong> (5.5) exist. Moreover, by (5.7) they are unique. Do<strong>in</strong>g so, we get, <strong>in</strong> the same way as <strong>in</strong> the<br />
proof of Proposition 4.3.1, the desired result. F<strong>in</strong>ally, we get the variation substitut<strong>in</strong>g the<br />
obta<strong>in</strong>ed values of A p,q at (5.6).<br />
An <strong>in</strong>terest<strong>in</strong>g particular case of the last theorem is the case of Lagrangian planes. This is<br />
the case stated by Alesker [Ale03] as a remarkable case of study. From the previous theorem<br />
we can give explicitly the constant <strong>in</strong> the theorem of Alesker reproduced at 2.3.5, but with<br />
respect to the Hermitian <strong>in</strong>tr<strong>in</strong>sic volumes def<strong>in</strong>ed by Bernig-Fu, and not directly by the bases<br />
def<strong>in</strong>ed by Alesker.<br />
Corollary 5.5.4. Let Ω be a regular doma<strong>in</strong> <strong>in</strong> C n with piecewise smooth boundary. Then,<br />
∫<br />
L R n<br />
χ(Ω ∩ L)dL = vol(G n,n,0)ω n<br />
n!<br />
⌊ n−1<br />
2 ⌋<br />
∑<br />
q=0<br />
where L R n denotes the <strong>space</strong> of Lagrangian planes <strong>in</strong> C n .<br />
( ) 2q − 1 −1<br />
4 q−n<br />
q − 1 2q + 1 µ n,q(Ω).<br />
5.6 Measure of Lagrangian planes <strong>in</strong> CK n (ɛ)<br />
Us<strong>in</strong>g the same techniques as <strong>in</strong> Chapter 4, it can be proved the follow<strong>in</strong>g result.<br />
Theorem 5.6.1. Let Ω ⊂ CK n (ɛ) be a regular doma<strong>in</strong>, X a smooth vector field def<strong>in</strong>ed at<br />
CK n (ɛ), φ t the flow associated to X and Ω t = φ t (Ω). Then,<br />
∫<br />
d<br />
dt∣ t=0<br />
L R n<br />
and<br />
if n is odd<br />
χ(Ω t ∩ L)dL = vol(G n,n,0 )ω n+1<br />
(n + 2)<br />
n!<br />
⌊ n−1<br />
2 ⌋ ∑<br />
q=0<br />
4 q−n+1<br />
2q + 3<br />
( ) 2q + 1 −1˜Γn−1,q (Ω),<br />
q + 1<br />
∫<br />
L R n<br />
χ(Ω ∩ L)dL = vol(G n,n,0)ω n<br />
n!<br />
n−1<br />
2∑<br />
q=0<br />
( ) 2q − 1 −1<br />
4 q−n<br />
q − 1 2q + 1 µ n,q(Ω), (5.8)<br />
and if n is even<br />
∫<br />
χ(Ω ∩ L)dL = vol(G n,n,0)<br />
· (5.9)<br />
L R n!<br />
n<br />
⎛<br />
n<br />
2∑<br />
( )<br />
· ⎝ 2q − 1 −1<br />
4 q−n n<br />
⎞<br />
ω n<br />
2∑<br />
( ) n −1<br />
q − 1 2q + 1 µ n,q(Ω) + ɛ i 2 −n+1 ω n−2i<br />
n<br />
2 + i µ<br />
n + 1 n+2i,<br />
n<br />
2 +i (Ω) ⎠ .<br />
q=0<br />
i=1<br />
Proof. In the same way as <strong>in</strong> the proof of Theorem 4.3.5, it is enough to prove that the<br />
variation <strong>in</strong> both sides co<strong>in</strong>cides.<br />
The variation on the left hand side of (5.9) and (5.8) co<strong>in</strong>cides and is <strong>in</strong>dependent on ɛ.<br />
Thus, it co<strong>in</strong>cides with the variation <strong>in</strong> (5.4).<br />
We compute the variation on the right hand side by us<strong>in</strong>g Proposition 4.1.7. Here, we just<br />
reproduce the computations when n is odd. For n even, a similar, but longer study can be<br />
done to verify expression (5.9).
5.6 Measure of Lagrangian planes <strong>in</strong> CK n (ɛ) 91<br />
Denote by E n (Ω) the right hand side of (5.8). Then, by (5.4) we have<br />
δ X E n (Ω)= vol(G n,n,0)ω n<br />
n!<br />
n−1<br />
2∑<br />
q=0<br />
( ) 2q − 1 −1<br />
4 q−n<br />
q − 1 2q + 1 2c n,n,q·<br />
(<br />
· c −1<br />
n,n−1,q (n − 2q)2˜Γn−1,q − c −1<br />
n,n−1,q−1 q2˜Γn−1,q−1<br />
+ c −1<br />
n,n−1,q−1 q(q + 1/2) ˜B n−1,q−1 − c n,n−1,q (n − 2q)(n − 2q − 1) ˜B n−1,q<br />
+ ɛ(c −1<br />
n,n+1,q+1 (n − 2q)(n − 2q − 1) ˜B n+1,q+1 − c −1<br />
n,n+1,q q(q + 1/2) ˜B n+1,q )<br />
n−3 (<br />
)<br />
= 2 vol(G n,n,0)ω n<br />
2∑ c n,n,q 4 q−n (n − 2q) 2<br />
{ {<br />
n!<br />
(2q + 1) ( )<br />
2q−1<br />
− c n,n,q+14 q−n+1 (q + 1) 2<br />
q=0<br />
q−1<br />
(2q + 3) ( )<br />
2q+1<br />
q<br />
(<br />
)<br />
c n,n,q+1 4 q−n+1 (q + 1)(q + 3/2)<br />
+<br />
(2q + 3) ( )<br />
2q+1<br />
− c n,n,q4 q−n (n − 2q)(n − 2q − 1)<br />
q<br />
(2q + 1) ( )<br />
2q−1<br />
q−1<br />
+ɛ<br />
(<br />
2∑ c n,n,q−1 4 q−n−1 (n − 2q + 2)(n − 2q + 1)<br />
(2q − 1) ( )<br />
2q−3<br />
− c n,n,q4 q−n q(q + 1 2 )<br />
)<br />
q−2<br />
(2q + 1) ( )<br />
2q−1<br />
q−1<br />
n−1<br />
q=1<br />
)<br />
c −1<br />
n,n−1,q ˜Γ n−1,q<br />
c −1<br />
n,n−1,q ˜B n−1,q }<br />
c −1<br />
n,n+1,q ˜B n+1,q }.<br />
In order to prove the result, it suffices to prove that this expression is <strong>in</strong>dependent on ɛ. As<br />
for ɛ = 0 we know that δ X E n (Ω) co<strong>in</strong>cides with (5.4), we get the result.<br />
Now, to prove the <strong>in</strong>dependence of ɛ, we collect the coefficient for each ˜B n−1,q and ˜B n+1,q ,<br />
and we prove that they vanish.
Appendix<br />
This appendix conta<strong>in</strong>s a constructive proof of Theorems 4.3.5 and 4.4.1.<br />
Proof of Theorem 4.3.5<br />
We prove that it is possible to f<strong>in</strong>d constants α k,q such that<br />
∫<br />
L C r<br />
χ(Ω ∩ L r )dL r = ∑ k,q<br />
⌊n/2⌋<br />
∑<br />
α k,q B k,q (Ω) + α 2j,j Γ 2j,j (Ω) + α 2n,n vol(Ω)<br />
j=1<br />
(A.10)<br />
where max{0, k − n} ≤ q < k/2 ≤ n.<br />
For ɛ = 0, the existence of these constants follows from the fact that Hermitian <strong>in</strong>tr<strong>in</strong>sic<br />
volumes constitute a basis of smooth valuations. If ɛ ≠ 0, we cannot ensure this fact. Anyway,<br />
we f<strong>in</strong>d the value of the previous constants impos<strong>in</strong>g that the variation <strong>in</strong> both sides of (A.10)<br />
co<strong>in</strong>cides. This is enough to prove (A.10). Indeed, take a deformation Ω t of Ω such that Ω t<br />
converges to a po<strong>in</strong>t. Then, both sides of (A.10) have the same variation and it vanishes <strong>in</strong><br />
the limit.<br />
The variation of the left hand side of (A.10) is given <strong>in</strong> Corollary 4.3.3. The variation of<br />
the right hand side can be computed us<strong>in</strong>g Proposition 4.1.7 and δ X vol = 2 ˜B 2n−1,n−1 .<br />
In the variation of the left hand side, just appear the terms { ˜B 2n−2r−1,q } q . Thus, the<br />
variation of the right hand side can only have these terms. On the other hand, the variation<br />
of a Hermitian <strong>in</strong>tr<strong>in</strong>sic volume B k,q with k even (resp. odd) has only terms ˜B a,b and ˜Γ a ′ ,b ′<br />
with a, a ′ odd (resp. even) (cf. Proposition 4.1.7). As the variation of the left hand side has<br />
only non-vanish<strong>in</strong>g terms with odd subscript, we just consider the valuations with first even<br />
subscript. Do<strong>in</strong>g also the change <strong>in</strong> (4.12), expression (A.10) reduces to<br />
∫<br />
L C r<br />
n−1<br />
∑<br />
(<br />
k=1<br />
n−1<br />
∑<br />
χ(Ω ∩ L r )dL r = (<br />
k=1<br />
∑k−1<br />
q=max{0,2k−n}<br />
C 2k,q B 2k,q ′ (Ω) + D 2k,kΓ ′ 2k,k (Ω)) + dvol(Ω). (A.11)<br />
Now, we start the study to f<strong>in</strong>d constants C k,q , D 2q,q , d such that<br />
∑k−1<br />
q=max{0,2k−n}<br />
⎛<br />
= vol(GC n−1,r )ω 2r+1(r + 1)<br />
)( ⎝<br />
n<br />
r)<br />
( n−1<br />
r<br />
C 2k,q δB 2k,q ′ (Ω) + D 2k,kδΓ ′ 2k,k (Ω) + dδvol(Ω)<br />
n−r−1<br />
∑<br />
q=max{0,n−2r−1}<br />
⎞<br />
( )<br />
2n − 2r − 2q − 1 1<br />
n − r − q 4 ˜B ′ n−r−q−1 2n−2r−1,q(Ω) ⎠ .<br />
By Proposition 4.1.7, this equation gives rise to a l<strong>in</strong>ear system. We write this l<strong>in</strong>ear system<br />
<strong>in</strong> matrix form Ax = b. Consider the vector of unknowns as<br />
x t = (C 2,0 , D 2,1 , C 4,0 , C 4,1 , D 4,2 , . . . , C 2c,max{0,2c−n} , . . . , D 2c,c , . . . , C 2n−2,n−2 , D 2n−2,n−1 , d).<br />
93
94 Appendix<br />
Vector b conta<strong>in</strong>s the coefficient of ˜B ′ k,q , ˜Γ ′ k,q<br />
given <strong>in</strong> (4.23), that is<br />
b t = vol(GC n−1,r ) ( ( )( ) ( )( n − r r + 1 n − r r + 1<br />
n! ( )<br />
n−1<br />
0, . . . , 0,<br />
,<br />
1 1 2 2<br />
r<br />
) 1<br />
2 , . . . , ( n − r<br />
r + 1<br />
)( r + 1<br />
r + 1<br />
) )<br />
r + 1<br />
4 r , 0, . . . , 0 .<br />
Note that b has all entries null except the ones correspond<strong>in</strong>g to ˜B 2n−2r−1,q ′ .<br />
The coefficients of the matrix A conta<strong>in</strong> the variation of each B<br />
2k,q ′ and Γ′ 2q,q with respect<br />
to ˜B r,s ′ and ˜Γ ′ r,s. We denote by (δB<br />
k,q ′ , ˜B r,s ), the coefficient of ˜Br,s <strong>in</strong> the variation of the<br />
valuation B<br />
k,q ′ . By Proposition 4.1.7<br />
(δB<br />
k,q ′ , ˜B r,s) ′ =<br />
(δB<br />
k,q ′ , ˜Γ ′ r,s) =<br />
(δΓ ′ 2q,q , ˜B r,s) ′ =<br />
(δΓ ′ 2q,q , ˜Γ ′ r,s) =<br />
⎧<br />
⎪⎨<br />
⎪⎩<br />
⎧<br />
⎨<br />
⎩<br />
⎧<br />
⎪⎨<br />
⎪⎩<br />
⎧<br />
⎨<br />
⎩<br />
2q(n + q − k + 1/2), if r = k − 1, s = q − 1<br />
−2(k − 2q)(k − 2q − 1), if r = k − 1, s = q<br />
2ɛ(k − 2q)(k − 2q − 1), if r = k + 1, s = q + 1<br />
−2ɛ(n − k + q)(q + 1/2), if r = k + 1, s = q<br />
0, otherwise.<br />
(k − 2q) 2 , if r = k − 1, s = q<br />
−(n + q − k)q, if r = k − 1, s = q − 1<br />
0, otherwise.<br />
4q(n − q + 1/2), if r = 2q − 1, s = q − 1<br />
−4ɛ((n − q)(2q + 3/2) − (q + 1)/2), if r = 2q + 1, s = q<br />
4ɛ 2 (n − q − 1)(q + 3/2), if r = 2q + 3, s = q + 1<br />
0, otherwise.<br />
−2(n − q)q, if r = 2q − 1, s = q − 1<br />
2ɛ(n − q − 1)(q + 1), if r = 2q + 1, s = q<br />
0, otherwise.<br />
Each column of the matrix A conta<strong>in</strong>s the variation of a valuation B<br />
2k,q ′ , Γ′ 2q,q or the volume.<br />
We take the valuations B<br />
2k,q ′ , Γ′ 2q,q <strong>in</strong> the same order as <strong>in</strong> the vector b. (The volume corresponds<br />
to the last column.) That is, the columns of A conta<strong>in</strong> the variation of the valuations<br />
<strong>in</strong> the follow<strong>in</strong>g order<br />
(δB ′ 2,0, δΓ ′ 2,1, δB ′ 4,0, δB ′ 4,1, δΓ ′ 4,2, . . . , δB ′ 2n−2,n−2, δΓ ′ 2n−2,n−1, δvol).<br />
We denote<br />
δB ′ 2k,· = (δB′ 2k,max{0,2k−n} , δB′ 2k,max{0,2k−n}+1 , . . . , δB′ 2k,k−1 ),<br />
δΓ ′ 2k,· = δΓ′ 2k,k ,<br />
˜B ′ 2k+1,· = ( ˜B ′ 2k+1,max 0,2k−n+1 , ˜B ′ 2k+1,max 0,2k−n+1+1 , . . . , ˜B ′ 2k+1,k )t ,<br />
˜Γ ′ 2k+1,· = (˜Γ ′ 2k+1,max 0,2k−n+1 , ˜Γ ′ 2k+1,max 0,2k−n+1+1 , . . . , ˜Γ ′ 2k+1,k )t .
Proof of Theorem 4.3.5 95<br />
Then, A has the follow<strong>in</strong>g boxes structure<br />
δB 2,· ′ δΓ ′ 2,· δB 4,· ′ δΓ ′ 4,· δB 6,· ′ δΓ ′ 6,· δB 8,· ′ δΓ ′ 8,· · · · δB 2n−4,· ′ δΓ ′ 2n−4,· δB 2n−2,· ′ δΓ ′ 2n−2,· δvol<br />
˜B 1,· ′ ∗ ∗<br />
˜Γ ′ 1,· ∗ ∗<br />
˜B 3,· ′ ∗ ɛ ∗ ɛ ∗ ∗<br />
˜Γ ′ 3,· ∗ ɛ ∗ ∗<br />
˜B 5,· ′ ∗ ɛ ∗ ɛ ∗ ɛ ∗ ∗<br />
˜Γ ′ 5,· ∗ ɛ ∗ ∗<br />
˜B 7,· ′ ∗ ɛ ∗ ɛ ∗ ɛ ∗ ∗<br />
˜Γ ′ 7,· ∗ ɛ ∗ ∗<br />
˜B 9,· ′ ∗ ɛ ∗ ɛ ∗ ɛ<br />
˜Γ ′ 9,·<br />
∗ ɛ<br />
.<br />
˜B 2n−3,· ′ ∗ ɛ ∗ ɛ ∗ ∗<br />
˜Γ ′ 2n−3,· ∗ ɛ ∗ ∗<br />
˜B 2n−1,· ′ ∗ ɛ ∗ ɛ ∗ ɛ ∗<br />
We denoted by ∗ the boxes of A with non-null coefficients and <strong>in</strong>dependent of ɛ, and by ∗ ɛ the<br />
boxes of A with non-null coefficients (for ɛ ≠ 0) and multiples of ɛ.<br />
The structure by boxes of the l<strong>in</strong>ear system given by A suggests the method of resolution:<br />
we start with the top box, and we get the value of variables C 2,q and D 2,1 . Then we solve the<br />
next bloc with rows ˜B ′ 3,·, ˜Γ ′ 3,·, us<strong>in</strong>g the value of variables C 2,q and D 2,1 . We can cont<strong>in</strong>ue this<br />
process, so that, once we know the value of variables C 2k,q and D 2k,k , we substitute it on the<br />
equations given by the rows ˜B ′ 2k+1,·, ˜Γ ′ 2k+1,·.<br />
Recall that the <strong>in</strong>dependent vector b has all terms null except the ones correspond<strong>in</strong>g to<br />
˜B ′ 2n−2r−1,q . Thus, the l<strong>in</strong>ear system is homogeneous for the first equations until ˜B′ 2n−2r−2,q ,<br />
and we can take C k,q = D 2q,q = 0 whenever k ≤ 2n − 2r − 1.<br />
By Theorem 4.3.1 we have a solution for the system Ax = b when ɛ = 0. This solution<br />
has C 2n−2r,q and D 2n−2r,n−r as non-null terms, and satisfies the equations until rows<br />
˜B 2n−2r,·, ′ ˜Γ ′ 2n−2r,· also for ɛ ≠ 0.<br />
So, we consider for all ɛ ∈ R and for all a ∈ {1, ..., m<strong>in</strong>{n − r, r}}<br />
C 2n−2r,n−r−a = vol(GC n−1,r )<br />
4 a n!<br />
D 2n−2r,n−r = vol(GC n−1,r )<br />
2 n!<br />
( ) n − 1 −1 ( n − r<br />
r a<br />
( ) n − 1 −1<br />
.<br />
r<br />
)( r<br />
a)<br />
,<br />
Now, we go on with the resolution of the l<strong>in</strong>ear system <strong>in</strong> CK n (ɛ). We study for each<br />
c ∈ {n − r + 1, . . . , n} the submatrix of A with all rows ˜B 2c−1,q ′ , ˜Γ ′ 2c−1,q . This matrix has the<br />
follow<strong>in</strong>g non-zero columns of A: δΓ ′ 2c−4,c−2 , δB′ 2c−2,·, δΓ′ 2c−2,c−1 , δB′ 2c,· and δΓ′ 2c,c .<br />
Suppose that we know the value of D 2c−4,c−2 , C 2c−2,q and D 2c−2,c−1 . Then, we can substitute<br />
them <strong>in</strong> the equations given by the rows correspond<strong>in</strong>g to ˜B 2c−1,q , ˜Γ 2c−1,q . We get<br />
equations with C 2c,q and D 2c,c as unknowns. If we denote i = max{0, 2c − n}, the matrix of<br />
the coefficients for the obta<strong>in</strong>ed equations (which corresponds to a matrix bloc of A <strong>in</strong>depen-
96 Appendix<br />
dent of ɛ) is<br />
δB 2c,i δB 2c,i+1 δB 2c,i+2 · · · δB 2c,c−2 δB 2c,c−1 δΓ 2c,c<br />
˜B 2c−1,i −4c(2c − 1) 2(n − 2c + 3/2)<br />
˜B 2c−1,i+1 −2(2c − 2)(2c − 3) 4(n − 2c + 5/2)<br />
˜B 2c−1,i+2 −2(2c − 4)(2c − 5)<br />
.<br />
˜B 2c−1,c−2 −24 2(c − 1)(n − c − 1 ) 2<br />
˜B 2c−1,c−1 −4<br />
˜Γ 2c−1,0 (2c) 2 −(n − 2c + 1)<br />
4c(n − c + 1 ) 2<br />
˜Γ 2c−1,1 (2c − 2) 2 −2(n − 2c + 2)<br />
˜Γ 2c−1,2 (2c − 4) 2<br />
.<br />
˜Γ 2c−1,c−2 4 2 −(c − 1)(n − 1)<br />
˜Γ 2c−1,c−1 4 −2c(n − c)<br />
The <strong>in</strong>dependent term is obta<strong>in</strong>ed from the <strong>in</strong>itial <strong>in</strong>dependent term b (which <strong>in</strong> these<br />
cases is always zero) and from the part of the <strong>in</strong>itial equation <strong>in</strong> which we substituted the<br />
value of D 2c−4,c−2 , C 2c−2,q , D 2c−2,c−1 . Compar<strong>in</strong>g the box structure of A on page 95 and its<br />
coefficients on page 94, we obta<strong>in</strong> that the <strong>in</strong>dependent term of this new l<strong>in</strong>ear system has<br />
zero the terms ˜Γ ′ 2c−1,q with q ∈ {max{0, 2c − n}, ..., c − 2}, and the term ˜Γ 2c−1,c−1 equals to<br />
2ɛ(n − c)cD 2c−2,c−1 .<br />
Now, we consider the equations given by rows ˜Γ ′ 2c−1,q and ˜B 2c−1,max{0,2c−n} <strong>in</strong> the previous<br />
matrix, which give a compatible l<strong>in</strong>ear system with one solution. The <strong>in</strong>dependent terms of<br />
the equation given by equation ˜B 2c−1,max{0,2c−n} is ɛ(n−2c+2)C 2c−2,max{0,2c−n−2} . The other<br />
ones are zero. Solv<strong>in</strong>g this system we get the variables C 2c,max{0,2c−n} and D 2c,c <strong>in</strong> terms of<br />
C 2c−2,max{0,2c−n−2} and D 2c−2,c−1 , which we suppose known.<br />
In order to avoid consider<strong>in</strong>g the maximum max{0, 2c − n − 2} we dist<strong>in</strong>guish two cases.<br />
First stage: 2c ≤ n. This case appears if 2r > n (s<strong>in</strong>ce c ∈ {n − r + 1, . . . , n}).<br />
The l<strong>in</strong>ear system we have to solve is given by the augmented matrix<br />
⎛<br />
⎞<br />
−4c(2c − 1) 2(n − 2c + 3/2) ɛ(n − 2c + 2)C 2c−2,0<br />
(2c) 2 −(n − 2c + 1)<br />
(2c − 2) 2 −2(n − 2c + 2)<br />
.<br />
⎜<br />
.. ⎟<br />
⎝<br />
⎠<br />
−(c − 1)(n − c − 1)<br />
4 −2c(n − c) −2ɛc(c − n)D 2c−2,c−1<br />
with variables {C 2c,0 , C 2c,1 , . . . , C 2c,c−1 , D 2c,c }.<br />
From the first two equations we obta<strong>in</strong><br />
ɛ(n − 2c + 2)(n − 2c + 1)<br />
C 2c,0 = C 2c−2,0 ,<br />
4c(n − c + 1)<br />
ɛ(n − 2c + 2)c<br />
C 2c,1 = C 2c−2,0 .<br />
(n − c + 1)<br />
For every q ∈ {0, ..., c − 2} the follow<strong>in</strong>g relations are satisfied<br />
and we get<br />
C 2c,q+1 =<br />
(2c − 2q) 2 C 2c,q = (q + 1)(n − 2c + q + 1)C 2c,q+1 ,<br />
4C 2c,c−1 − 2c(n − c)D 2c,c = −2ɛc(n − c)D 2c−2,c−1 ,<br />
ɛ4 q (n − 2c + 2)!c!(c − 1)!<br />
(n − c + 1)(q + 1)!(n − 2c + q + 1)!(c − q − 1)!(c − q − 1)! C 2c−2,0,<br />
D 2c,c = ɛ(D 2c−2,c−1 +<br />
2(n − 2c + 2)!(c − 1)!4c−2<br />
C 2c−2,0 ).<br />
(n − c + 1)!<br />
(A.12)
Proof of Theorem 4.3.5 97<br />
As the known constants are C 2n−2r,q and D 2n−2r,n−r we write the previous ones <strong>in</strong> terms<br />
of these, us<strong>in</strong>g the recurrence we just obta<strong>in</strong>ed.<br />
C 2c,0 =<br />
ɛ(n − 2c + 2)(n − 2c + 1)<br />
C 2c−2,0<br />
4c(n − c + 1)<br />
= ɛc−(n−r) (n − 2c + 2)(n − 2c + 1) · ... · (n − 2n + 2r − 2 + 2)(n − 2n + 2r − 2 + 1)<br />
4 c−(n−r) c(c − 1) · ... · (n − r + 1)(n − c + 1)(n − c + 2) · ... · (n − (n − r))<br />
(A.13)<br />
= ɛc−(n−r) (2r − n)!(n − r)!(n − c)!vol(G C n−1,r ) ( ) n − 1 −1 ( ) r<br />
4 c−(n−r) (n − 2c)!c!r!4 n−r n!<br />
r n − r<br />
= ɛc−(n−r) vol(G C n−1,r ) ( ) n − 1 −1 ( ) n − c<br />
4 c ,<br />
n!<br />
r c<br />
C 2c,q+1 =<br />
C 2n−2r,0<br />
ɛ(n − 2c + 2)4 q (n − 2c + 1)!c!(c − 1)!<br />
(n − c + 1)(q + 1)!(n − 2c + q + 1)!(c − q − 1)!(c − q − 1)! C 2c−2,0 (A.14)<br />
= ɛ(n − 2c + 2)4q (n − 2c + 1)!c!(c − 1)!ɛ c−1−(n−r) vol(G C n−1,r )(n − c + 1)!<br />
(n − c + 1)(q + 1)!(n − 2c + q + 1)!((c − q − 1)!) 2 4 c−1 n!(c − 1)!(n − 2c + 2)!<br />
ɛ c−(n−r) c!(n − c)!vol(G C n−1,r<br />
=<br />
)<br />
( ) n − 1 −1<br />
4 c−q−1 (q + 1)!(n − 2c + q + 1)!(c − q − 1)!(c − q − 1)!n! r<br />
= ɛc−(n−r) vol(G C n−1,r ) ( ) n − 1 −1 ( )( )<br />
c n − c<br />
4 c−q−1 ,<br />
n! r q + 1 c − q − 1<br />
( n − 1<br />
)<br />
2(n − 2c + 2)!(c − 1)!4c−2<br />
D 2c,c = ɛ<br />
(D 2c−2,c−1 + C 2c−2,0<br />
(n − c + 1)!<br />
(<br />
2(n − 2c + 2)!(c − 1)!4c−2 ɛ c−1−(n−r) vol(G C ( ) )<br />
n−1,r )(n − c + 1)! n − 1 −1<br />
= ɛ D 2c−2,c−1 +<br />
(n − c + 1)! 4 c−1 n!(c − 1)!(n − 2c + 2)! r<br />
= ɛD 2c−2,c−1 + ɛc−(n−r) vol(G C n−1,r ) ( ) n − 1 −1<br />
2 n!<br />
r<br />
= ɛ c−(n−r) D 2n−2r,n−r + (c − (n − r)) ɛc−(n−r) vol(G C n−1,r ) ( ) n − 1 −1<br />
(A.15)<br />
2 n!<br />
r<br />
= ɛc−(n−r) vol(G C n−1,r ) ( ) n − 1 −1<br />
+ (c − (n − r)) ɛc−(n−r) vol(G C n−1,r ) ( ) n − 1 −1<br />
2 n!<br />
r<br />
2 n!<br />
r<br />
= ɛc−(n−r) vol(G C n−1,r )<br />
( ) n − 1 −1<br />
(c + r − n + 1) .<br />
2 n!<br />
r<br />
Thus, we get the value of the unknowns (<strong>in</strong> the vector x) until position D 2⌊n/2⌋,⌊n/2⌋ .<br />
Second stage: 2c ≥ n. Note that B 2c,q ′ is def<strong>in</strong>ed if q ≥ 2c − n > 0. In this case, 2c ≥ n,<br />
and the system we have to solve has the same structure as <strong>in</strong> the previous case (2c ≤ n) but<br />
with less equations and unknowns. Tak<strong>in</strong>g the same equations as <strong>in</strong> the previous case 2c ≤ n,<br />
we obta<strong>in</strong>, as augmented matrix,<br />
⎛<br />
(2c − n) ɛ((4c − 2n − 1)C 2c−2,2c−n−1−<br />
−4(n − c + 1)(2n − 2c + 1)C 2c−2,2c−n−2)<br />
⎜ (2n − 2c)<br />
⎝<br />
2 −(2c − n − 1)<br />
⎟<br />
⎠ .<br />
4 −2c(n − c) −2ɛc(c − n)D 2c−2,c−1<br />
⎞<br />
r<br />
) −1
98 Appendix<br />
From the first equation, it follows<br />
ɛ<br />
C 2c,2c−n =<br />
2c − n ((4c−2n−1)C 2c−2,2c−n−1 −2(2n−2c+2)(2n−2c+1)C 2c−2,2c−n−2 ). (A.16)<br />
For a ∈ {0, ..., n − c}, by relation<br />
we get<br />
and from<br />
C 2c,2c−n+a =<br />
and (A.17) we get<br />
(2n − 2c − 2a + 2) 2 C 2c,2c−n+a−1 = a(2c − n + a)C 2c,2c−n+a<br />
=<br />
=<br />
4(n − c − a + 1)2<br />
C 2c,2c−n+a−1<br />
a(2c − n + a)<br />
4 a (n − c − a + 1) 2 (n − c − a + 2) 2 . . . (n − c) 2<br />
a(a − 1) . . . 2(2c − n + a)(2c − n + a − 1) . . . (2c − n + 1) C 2c,2c−n<br />
4 a (n − c)!(n − c)!(2c − n)!<br />
a!(n − c − a)!(n − c − a)!(2c − n + a)! C 2c,2c−n<br />
4C 2c,c−1 − 2(n − c)cD 2c,c = −2ɛ(n − c)(c)D 2c−2,c−1<br />
D 2c,c = ɛD 2c−2,c−1 +<br />
2<br />
c(n − c) C 2c,c−1<br />
= ɛD 2c−2,c−1 + 2 · 4n−c−1 (n − c)!(n − c)!(2c − n)!<br />
C 2c,2c−n<br />
c(n − c)(n − c − 1)!(c − 1)!<br />
= ɛD 2c−2,c−1 + 2 4n−c−1 (n − c)!(2c − n)!<br />
C 2c,2c−n .<br />
c!<br />
(A.17)<br />
(A.18)<br />
In order to obta<strong>in</strong> the value of C 2c,2c−n we use the value of C 2c−2,2c−n−1 and C 2c−2,2c−n−2<br />
if c ∈ {n − r, ..., ⌊n/2⌋}. We consider c 0 = ⌊ n+2<br />
2 ⌋.<br />
From the previous case 2c ≤ n we know the value of the unknowns C 2c0 −2,2c 0 −n−1,<br />
C 2c0 −2,2c 0 −n−2 and D 2c0 −2,c 0 −1. For n even we have (we omit the analogous computation<br />
for n odd)<br />
C 2c0 −2,2c 0 −n−1 = C n,1 = ɛn/2−(n−r) vol(G C n−1,r )(n/2)!(n/2)! ( ) n − 1 −1<br />
4 n/2−1 (A.19)<br />
n!((n − 2)/2)!((n − 2)/2)! r<br />
= ɛr−n/2 vol(G C n−1,r ) ( ) n − 1 −1<br />
2 n n 2 ,<br />
n!<br />
r<br />
C 2c0 −2,2c 0 −n−2 = C n,0 = ɛr−n/2 vol(G C n−1,r )<br />
2 n n!<br />
( ) n − 1 −1<br />
,<br />
r<br />
D 2c0 −2,c 0 −1 = D n,n/2 = ɛn/2−(n−r) vol(G C n−1,r )<br />
(r − n ( ) n − 1 −1<br />
2 n!<br />
2 + 1) .<br />
r<br />
Then<br />
ɛ<br />
C n+2,2 =<br />
2c − n ((4c − 2n − 1)C 2c−2,2c−n−1 − 2(2n − 2c + 2)(2n − 2c + 1)C 2c−2,2c−n−2 )<br />
= ɛr−n/2+1 vol(G C n−1,r )<br />
( ) n − 1 −1<br />
2 n+1 (3n 2 − 2n(n − 1))<br />
n!<br />
r<br />
= ɛr−n/2+1 vol(G C n−1,r ) ( ) n − 1 −1<br />
2 n+1 n(n + 2) .<br />
n!<br />
r
Proof of Theorem 4.3.5 99<br />
Once we know the expression of C n+2,2 we f<strong>in</strong>d the value of C 2c,2c−n , for every c ∈ {⌊n/2⌋, . . . , n},<br />
us<strong>in</strong>g the recurrence for C 2c,2c−n and C 2c−2,2c−n−1 . First, we have<br />
(A.16)<br />
C 2c,2c−n =<br />
(A.17)<br />
ɛ<br />
2c − n ((4c − 2n − 1)C 2c−2,2c−n−1 − 2(2n − 2c + 2)(2n − 2c + 1)C 2c−2,2c−n−2 )<br />
= ɛC 2c−2,2c−n−2<br />
·<br />
(<br />
2c − n<br />
(4c − 2n − 1)4(n − c + 1)!(n − c + 1)!(2c − n − 2)!<br />
·<br />
(n − c)!(n − c)!(2c − n − 1)!<br />
(<br />
4ɛ(n − c + 1) (4c − 2n − 1)(n − c + 1) − (2n − 2c + 1)(2c − n − 1)<br />
=<br />
2c − n<br />
(2c − n − 1)<br />
4ɛ(n − c + 1)c<br />
=<br />
(2c − n)(2c − n − 1) C 2c−2,2c−n−2.<br />
)<br />
− 4(n − c + 1)(2n − 2c + 1)<br />
)<br />
C 2c−2,2c−n−2<br />
We go on with this recurrence until C ∗,∗−n with ∗ ≤ n+2<br />
2<br />
. In this case we know the value of<br />
the constants, and we can f<strong>in</strong>d the value of C 2c,2c−n .<br />
C 2c,2c−n = (4ɛ)c−(n+2)/2 c(c − 1) · ... · ((n + 4)/2)(n − c + 1)(n − c + 2) · ... · (n/2 − 1)<br />
C n+2,2<br />
(2c − n)(2c − n − 1) · ... · 4 · 3<br />
F<strong>in</strong>ally,<br />
= (4ɛ)c−(n+2)/2 c!((n − 2)/2)!2<br />
((n + 2)/2)!(n − c)!(2c − n)! C n+2,2<br />
= (4ɛ)c−(n+2)/2 2 3 ( ) c ɛ r−n/2+1 vol(G C n−1,r ) ( ) n − 1 −1<br />
(n + 2)n 2c − n 2 n+1 n(n + 2)<br />
n!<br />
r<br />
= ɛc−(n−r) vol(G C n−1,r ) ( ) n − 1 −1 ( ) c<br />
4 n−c . (A.20)<br />
n! r 2c − n<br />
and<br />
4 a (n − c)!(n − c)!(2c − n)! ɛ c−(n−r) c!vol(G C n−1,r<br />
C 2c,2c−n+a =<br />
) ( n − 1<br />
a!(n − c − a)!(n − c − a)!(2c − n + a)! 4 n−c (2c − n)!(n − c)!n! r<br />
= ɛc−(n−r) vol(G C n−1,r ) ( ) n − 1 −1 ( )( )<br />
n − c c<br />
4 n−c−a n! r n − c − a 2c − n + a<br />
) −1<br />
= ɛD 2c−2,c−1 + 2 4n−c−1 (n − c)!(2c − n)! ɛ c−(n−r) c!vol(G C n−1,r ) ( n − 1<br />
c! 4 n−c (2c − n)!(n − c)!n! r<br />
= ɛD 2c−2,c−1 + ɛc−(n−r) vol(G C n−1,r ) ( ) n − 1 −1<br />
2 n!<br />
r<br />
= ɛ c−n/2 D n,n/2 + (c − n/2) ɛc−(n−r) vol(G C n−1,r ) ( ) n − 1 −1<br />
2 n!<br />
r<br />
= ɛc−(n−r) vol(G C n−1,r )<br />
( ) n − 1 −1<br />
(c + r − n + 1) .<br />
2 n!<br />
r<br />
D 2c,c<br />
(A.17) and (A.20)<br />
) −1<br />
To determ<strong>in</strong>e the value of d, the coefficient of δvol, we consider the last equation of the
100 Appendix<br />
<strong>in</strong>itial l<strong>in</strong>ear system<br />
So,<br />
d<br />
(n − 1)! = −2ɛ2 (2n − 1)D 2n−4,n−2 − 4ɛC 2n−2,n−2 + 2ɛ(3n − 1)D 2n−2,n−1<br />
= 2ɛ r (−(2n − 1)(r − 1) − (n − 1) + (3n − 1)r) vol(GC n−1,r ) ( n − 1<br />
2 n! r<br />
= ɛr vol(G C n−1,r ) ( ) n − 1 −1<br />
n(r + 1) .<br />
n!<br />
r<br />
( ) n − 1 −1<br />
d = ɛ r vol(G C n−1,r)(r + 1) .<br />
r<br />
Now, we have to prove that the given solution satisfies all the equations we did not use to<br />
solve the system. This is because we cannot ensure that the equation (A.10) has solution.<br />
Let us study first the case 2c ≤ n. Consider the matrix on page 96. The rows we did not<br />
use correspond to ˜B 2c−1,q ′ , q ∈ {1, ..., c − 1}. Suppose q ≠ c − 1. The equations given by this<br />
row are ˜B 2c−1,q<br />
′<br />
−(2c − 2q)(2c − 2q − q)C 2c,q + (q + 1)(n − 2c + q + 3/2)C 2c,q+1<br />
= −ɛ(2c − 2q)(2c − 2q − 1)C 2c−2,q−1 + ɛ(n − 2c + q + 2)(q + 1/2)C 2c−2,q .<br />
For q = c − 1, the equation is<br />
2ɛ 2 (n − c + 1)(c − 1/2)D 2c−4,c−2 + 2ɛC 2c−2,c−2 − 2ɛ((n − c + 1)(2c − 1/2) − c/2)D 2c−2,c−1<br />
= −2C 2c,c−1 + 2c(n − 2 + 1/2)D 2c,c .<br />
Substitut<strong>in</strong>g the value of each C ∗,· and D ∗,· given on page 97 we prove that the equations are<br />
satisfied.<br />
In the same way, we can prove that all equations appear<strong>in</strong>g <strong>in</strong> the case 2c ≥ n are also<br />
satisfied.<br />
F<strong>in</strong>ally, us<strong>in</strong>g aga<strong>in</strong> the relation <strong>in</strong> (4.12), we get the result with respect to {B k,q , Γ k,q }.<br />
Proof of Theorem 4.4.1<br />
The idea of the proof of this theorem is the same as for Theorem 4.3.5. In the same way, if the<br />
variation δ X is the same <strong>in</strong> both sides, for all differentiable vector field X, then the expression<br />
holds.<br />
From the Gauss-Bonnet-Chern formula, we know that χ(Ω) can be written as the <strong>in</strong>tegral<br />
over N(Ω) of a differential form O(2n)-<strong>in</strong>variant, and also U(n)-<strong>in</strong>variant. Thus, by Proposition<br />
2.4.5 there exist constants C k,q , D k,q , d such that<br />
) −1<br />
χ(Ω) = ∑ k,q<br />
⌊ n 2 ⌋<br />
C k,q B k,q ′ (Ω) + ∑<br />
D 2j,j Γ ′ 2j,j(Ω) + dvol(Ω)<br />
j=1<br />
(A.21)<br />
where max{0, k − n} ≤ q < k/2 ≤ n, and B<br />
k,q ′ and Γ′ k,q<br />
are the valuations def<strong>in</strong>ed <strong>in</strong> (4.12).<br />
Tak<strong>in</strong>g the variation <strong>in</strong> both sides of the previous equality we have<br />
0 = ∑ (c k,q ˜B′ k,q (Ω) + d 2q,q ˜Γ′ 2q,q (Ω))<br />
k,q<br />
with c k,q and d k,q l<strong>in</strong>ear comb<strong>in</strong>ation of C k,q and D 2q,q .
Proof of Theorem 4.4.1 101<br />
Thus, we have to impose c k,q = d k,q = 0.<br />
The variation of Γ ′ 0,0 <strong>in</strong> CKn (ɛ) is (cf. Corollary 4.1.9)<br />
δΓ ′ 0,0(Ω) = 2ɛ(−(3n − 1) ˜B ′ 1,0(Ω) + (n − 1)˜Γ ′ 1,0(Ω) + 3ɛ(n − 1) ˜B ′ 3,1(Ω)).<br />
It is necessary to cancel the variation of the terms ˜B 1,0 ′ , ˜B 3,1 ′ and ˜Γ ′ 1,0 . By Proposition 4.1.7 we<br />
have that the variation of a valuation B<br />
k,q ′ <strong>in</strong> CKn (ɛ) with k even (resp. odd) has only terms<br />
˜B<br />
k ′ ′ ,q<br />
and ˜Γ ′ ′ k ′ ,q<br />
with k ′ odd (resp. even). Thus, <strong>in</strong> the expresssion (A.21) we can restrict the<br />
′<br />
value of k to k even, and (A.21) can be reduced to<br />
⎛<br />
n−1<br />
∑<br />
χ(Ω) = ⎝<br />
k=0<br />
∑k−1<br />
q=max{0,2k−n}<br />
C 2k,q B ′ 2k,q (Ω) + D 2k,kΓ ′ 2k,k (Ω) ⎞<br />
⎠ + dvol(Ω).<br />
(A.22)<br />
The right hand side <strong>in</strong> the previous equality co<strong>in</strong>cides with the right hand side of (A.11) plus<br />
the term D 0,0 Γ ′ 0,0 . Thus, the variation is very similar and the l<strong>in</strong>ear system we have to solve<br />
will be also very similar to the one solved <strong>in</strong> Theorem 4.3.5. The only different equations are<br />
the ones given by c 1,0 = 0, d 1,0 = 0 and c 3,1 = 0, that is<br />
−ɛ(3n − 1)D 0,0 − 2C 2,0 + 2(n − 1/2)D 2,1 = 0,<br />
ɛ(n − 1)D 0,0 + 2C 2,0 − (n − 1)D 2,1 = 0,<br />
3ɛ 2 (n − 1)D 0,0 + 2ɛC 2,0 − ɛ(7n − 9)D 2,1 − 2C 4,1 + 2(2n − 3)D 4,2 = 0.<br />
(A.23)<br />
We f<strong>in</strong>d the value of D 0,0 for ɛ = 0, i.e. <strong>in</strong> C n , us<strong>in</strong>g the Gauss-Bonnet formula so that<br />
D 0,0 =<br />
1<br />
O 2n−1 (n − 1)! = 1<br />
2nω 2n (n − 1)! = n!<br />
2 n!π n = 1<br />
2π n .<br />
The choice for the value of D 0,0 ensures that both sides <strong>in</strong> (A.21) co<strong>in</strong>cide when Ω collapses<br />
to a po<strong>in</strong>t.<br />
From the first two equation and the value of D 0,0 we get<br />
C 2,0 = ɛ 2 (n − 1)D 0,0 =<br />
D 2,1 = 2ɛD 0,0 = ɛ<br />
π n .<br />
ɛ(n − 1)<br />
4π n ,<br />
In order to f<strong>in</strong>d the value of C 4,1 and D 4,2 we consider the equations given by {c 3,0 =<br />
0, c 3,1 = 0, d 3,0 = 0, d 3,1 = 0}. The equation c 3,1 = 0 is the one given <strong>in</strong> (A.23), and the others<br />
are<br />
−ɛ(n − 2)C 2,0 − 24C 4,0 + (2n − 5)C 4,1 = 0,<br />
16C 4,0 − (n − 3)C 4,1 = 0,<br />
ɛ(n − 2)D 2,1 + C 4,1 − (n − 2)D 4,2 = 0.<br />
(Note that they co<strong>in</strong>cide with the ones <strong>in</strong> Theorem 4.3.5.) Solv<strong>in</strong>g the system given by these<br />
3 equations and (A.23), we get that it is compatible with solution<br />
C 4,0 =<br />
ɛ2<br />
32π n (n − 2)(n − 3), C 4,1 = ɛ2<br />
(n − 2),<br />
2πn D<br />
4,2 = 3ɛ2<br />
2π n .<br />
To f<strong>in</strong>d the value of the unknowns C 2c,q , D 2c,c with c ≥ 3, we have to solve the same<br />
equations as <strong>in</strong> the proof of Theorem 4.3.5. We can use the same relations if we first prove<br />
that C 4,0 , C 4,1 and D 4,2 also satisfies (A.12). We have to check it because variables C 4,0 , C 4,1
102 Appendix<br />
and D 4,2 here were obta<strong>in</strong>ed solv<strong>in</strong>g another l<strong>in</strong>ear system. But, it is straightforward verified.<br />
So, we get the same relation among the unknowns.<br />
Thus, from the equalities (A.13), (A.14), (A.15), (A.17) and (A.18), with r = n − 1, and<br />
the computation of C n,1 , C n,0 , D n,⌊n/2⌋ <strong>in</strong> the same way as <strong>in</strong> (A.19) we have<br />
χ(Ω) =<br />
n−1<br />
∑<br />
c=0<br />
c=0<br />
⎛<br />
ɛ c<br />
⎝<br />
π n<br />
∑c−1<br />
q=max{0,2c−n}<br />
1<br />
4 c−q ( c<br />
q<br />
)( n − c<br />
c − q<br />
)<br />
B ′ 2c,q(Ω) + c + 1<br />
Us<strong>in</strong>g the relation <strong>in</strong> (4.12) we get the stated expression<br />
⎛<br />
n−1<br />
∑ ɛ c ∑c−1<br />
( )( )<br />
χ(Ω) = ⎝<br />
1 c n − c<br />
π n 4 c−q B ′<br />
q c − q<br />
2c,q(Ω) + c + 1<br />
=<br />
=<br />
=<br />
n−1<br />
∑<br />
c=0<br />
+<br />
n−1<br />
∑<br />
c=0<br />
n−1<br />
∑<br />
c=0<br />
ɛ c (<br />
π n<br />
q=max{0,2c−n}<br />
∑c−1<br />
q=max{0,2c−n}<br />
2(c + 1)<br />
2<br />
ɛ c (<br />
π n<br />
2 Γ′ 2c,c(Ω)<br />
2 Γ′ 2c,c(Ω)<br />
⎞<br />
⎠+ ɛn (n + 1)!<br />
vol(Ω).<br />
⎞<br />
π n<br />
⎠ + ɛn (n + 1)!<br />
vol(Ω)<br />
1 c!(n − c)!q!(n − 2c + q)!(2c − 2q)!ω 2n−2c<br />
4 c−q B 2c,q (Ω)+<br />
q!(c − q)!(c − q)!(n − 2c + q)!<br />
c!(n − 2c + c)!(2c − 2c)!ω 2n−2c Γ 2c,c (Ω) ) + ɛn (n + 1)!<br />
vol(Ω)<br />
∑c−1<br />
q=max{0,2c−n}<br />
( )<br />
1 2c − 2q c!(n − c)!π<br />
n−c<br />
4 c−q B 2c,q (Ω)<br />
c − q (n − c)!<br />
π n−c<br />
+ (c + 1)!(n − c)!<br />
(n − c)! Γ 2c,c(Ω) ) + ɛn (n + 1)!<br />
π n vol(Ω)<br />
⎛<br />
ɛ c<br />
π c c! ∑c−1<br />
( )<br />
⎝<br />
1 2c − 2q<br />
4 c−q c − q<br />
q=max{0,2c−n}<br />
π n<br />
π n<br />
B 2c,q (Ω) + (c + 1)Γ 2c,c (Ω) ⎠ + ɛn (n + 1)!<br />
vol(Ω).<br />
⎞<br />
π n
Bibliography<br />
[AB08] S. Alesker and A. Bernig. The product on smooth and generalized valuations.<br />
arXiv:0904.1347v1, 2008.<br />
[Aba]<br />
J. Abardia. Average of mean curvature <strong>in</strong>tegral <strong>in</strong> <strong>complex</strong> <strong>space</strong> <strong>forms</strong>. To appear<br />
<strong>in</strong> Advances <strong>in</strong> Geometry.<br />
[AF08] S. Alesker and J.H.G. Fu. Theory of valuations on manifolds. III: Multiplicative<br />
structure <strong>in</strong> the general case. Trans. Am. Math. Soc., 360(4):1951–1981, 2008.<br />
[AGS09] J. Abardia, E. Gallego, and G. Solanes. Gauss-bonnet theorem and crofton type formulas<br />
<strong>in</strong> <strong>complex</strong> <strong>space</strong> <strong>forms</strong>. Prepr<strong>in</strong>ts de la Universitat Autònoma de Barcelona,<br />
2009.<br />
[Ale01]<br />
[Ale03]<br />
[Ale04]<br />
S. Alesker. Description of translation <strong>in</strong>variant valuations on convex sets with solution<br />
of P. McMullen’s conjecture. Geom. Funct. Anal., 11(2):244–272, 2001.<br />
S. Alesker. Hard Lefschetz theorem for valuations, <strong>complex</strong> <strong>in</strong>tegral <strong>geometry</strong>, and<br />
unitarily <strong>in</strong>variant valuations. J. Differential Geom., 63(1):63–95, 2003.<br />
S. Alesker. SU (2)-<strong>in</strong>variant valuations. Milman, V. D. (ed.) et al., Geometric aspects<br />
of functional analysis. Papers from the Israel sem<strong>in</strong>ar (GAFA) 2002–2003. Berl<strong>in</strong>:<br />
Spr<strong>in</strong>ger. Lecture Notes <strong>in</strong> Mathematics 1850, 21-29 (2004)., 2004.<br />
[Ale06a] S. Alesker. Theory of valuations on manifolds. I: L<strong>in</strong>ear <strong>space</strong>s. Isr. J. Math.,<br />
156:311–339, 2006.<br />
[Ale06b] S. Alesker. Theory of valuations on manifolds. II. Adv. Math., 207(1):420–454, 2006.<br />
[Ale07a] S. Alesker. Theory of valuations on manifolds: a survey. Geom. Funct. Anal.,<br />
17(4):1321–1341, 2007.<br />
[Ale07b] S. Alesker. Theory of valuations on manifolds. IV: New properties of the multiplicative<br />
structure. Milman, Vitali D. (ed.) et al., Geometric aspects of functional<br />
analysis. Proceed<strong>in</strong>gs of the Israel sem<strong>in</strong>ar (GFA) 2004-2005. Berl<strong>in</strong>: Spr<strong>in</strong>ger. Lecture<br />
Notes <strong>in</strong> Mathematics 1910, 1-44 (2007)., 2007.<br />
[ÁPF04] J. C. Álvarez Paiva and E. Fernandes. What is a Crofton formula<br />
42:95–108, 2003/04.<br />
Math. Notae,<br />
[Ber08a] A. Bernig. A hadwiger-type theorem for the special unitary group. Geometric and<br />
Functional Analysis (to appear), 2008.<br />
[Ber08b] A. Bernig. <strong>Integral</strong> <strong>geometry</strong> under G 2 and Sp<strong>in</strong>(7). arXiv:0803.3885v1, 2008.<br />
[BF08] A. Bernig and J.H.G. Fu. Hermitian <strong>in</strong>tegral <strong>geometry</strong>. arXiv:0801.0711v5, 2008.<br />
103
104 Bibliography<br />
[Bla55]<br />
[Bla76]<br />
[Bor49]<br />
W. Blaschke. Vorlesungen über <strong>Integral</strong>geometrie. 3. Aufl. Berl<strong>in</strong>: VEB Deutscher<br />
Verlag der Wissenschaften 130 S., 44 Fig. im Text. , 1955.<br />
D.E. Blair. Contact manifolds <strong>in</strong> Riemannian <strong>geometry</strong>. Spr<strong>in</strong>ger-Verlag, Berl<strong>in</strong>,<br />
1976. Lecture Notes <strong>in</strong> Mathematics, Vol. 509.<br />
A. Borel. Some remarks about Lie groups transitive on spheres and tori. Bull. Amer.<br />
Math. Soc., 55:580–587, 1949.<br />
[Bor50] A. Borel. Le plan projectif des octaves et les sphères comme e<strong>space</strong>s homogènes. C.<br />
R. Acad. Sci. Paris, 230:1378–1380, 1950.<br />
[GG00]<br />
[Gol99]<br />
C. Gorodski and N. Gusevskii. Complete m<strong>in</strong>imal hypersurfaces <strong>in</strong> <strong>complex</strong> hyperbolic<br />
<strong>space</strong>. Manuscripta Math., 103(2):221–240, 2000.<br />
W. M. Goldman. Complex hyperbolic <strong>geometry</strong>. Oxford Mathematical Monographs.<br />
The Clarendon Press Oxford University Press, New York, 1999. Oxford Science<br />
Publications.<br />
[Gra73] A. Gray. The volume of a small geodesic ball of a Riemannian manifold. Mich.<br />
Math. J., 20:329–344, 1973.<br />
[Gri78]<br />
P.A. Griffiths. Complex differential and <strong>in</strong>tegral <strong>geometry</strong> and curvature <strong>in</strong>tegrals<br />
associated to s<strong>in</strong>gularities of <strong>complex</strong> analytic varieties. Duke Math. J., 45:427–512,<br />
1978.<br />
[Had57] H. Hadwiger. Vorlesungen über Inhalt, Oberfläche und Isoperimetrie. Spr<strong>in</strong>ger-<br />
Verlag, Berl<strong>in</strong>, 1957.<br />
[Hsi98] P.H. Hsieh. L<strong>in</strong>ear submanifolds and bisectors <strong>in</strong> CH n . Forum Math., 10(4):413–434,<br />
1998.<br />
[KN69] S. Kobayashi and K. Nomizu. Foundations of differential <strong>geometry</strong>. Vol. II. Interscience<br />
Tracts <strong>in</strong> Pure and Applied Mathematics, No. 15 Vol. II. Interscience<br />
Publishers John Wiley & Sons, Inc., New York-London-Sydney, 1969.<br />
[KR97]<br />
[LS82]<br />
D.A. Kla<strong>in</strong> and G.-C. Rota. Introduction to geometric probability. Lezioni L<strong>in</strong>cee.<br />
Cambridge: Cambridge University Press. Rome: Accademia Nazionale dei L<strong>in</strong>cei,<br />
xiv, 178, 1997.<br />
R. Langev<strong>in</strong> and T. Shifr<strong>in</strong>. Polar varieties and <strong>in</strong>tegral <strong>geometry</strong>. Amer. J. Math.,<br />
104(3):553–605, 1982.<br />
[McM77] P. McMullen. Valuations and Euler-type relations on certa<strong>in</strong> classes of convex polytopes.<br />
Proc. London Math. Soc. (3), 35(1):113–135, 1977.<br />
[Mon85]<br />
[MS43]<br />
[MS83]<br />
[O’N83]<br />
S. Montiel. Real hypersurfaces of a <strong>complex</strong> hyperbolic <strong>space</strong>. J. Math. Soc. Japan,<br />
37(3):515–535, 1985.<br />
D. Montgomery and H. Samelson. Transformation groups of spheres. Ann. of Math.<br />
(2), 44:454–470, 1943.<br />
P. McMullen and R. Schneider. Valuations on convex bodies. In Convexity and its<br />
applications, pages 170–247. Birkhäuser, Basel, 1983.<br />
B. O’Neill. Semi-Riemannian <strong>geometry</strong>, volume 103 of Pure and Applied Mathematics.<br />
Academic Press Inc. [Harcourt Brace Jovanovich Publishers], New York, 1983.<br />
With applications to relativity.
Bibliography 105<br />
[Par02]<br />
H. Park. K<strong>in</strong>ematic formulas for the real sub<strong>space</strong>s of <strong>complex</strong> <strong>space</strong> <strong>forms</strong> of dimension<br />
2 and 3. PhD-thesis. University of Georgia, 2002.<br />
[Rum94] M. Rum<strong>in</strong>. Formes différentielles sur les variétés de contact. J. Differential Geom.,<br />
39(2):281–330, 1994.<br />
[San52] L. A. Santaló. <strong>Integral</strong> <strong>geometry</strong> <strong>in</strong> Hermitian <strong>space</strong>s. Amer. J. Math., 74:423–434,<br />
1952.<br />
[San04]<br />
[Sch93]<br />
[Sol06]<br />
[War71]<br />
L. A. Santaló. <strong>Integral</strong> <strong>geometry</strong> and geometric probability. Cambridge Mathematical<br />
Library. Cambridge University Press, Cambridge, second edition, 2004. With a<br />
foreword by Mark Kac.<br />
R. Schneider. Convex bodies: the Brunn-M<strong>in</strong>kowski theory, volume 44 of Encyclopedia<br />
of Mathematics and its Applications. Cambridge University Press, Cambridge,<br />
1993.<br />
G. Solanes. <strong>Integral</strong> <strong>geometry</strong> and the Gauss-Bonnet theorem <strong>in</strong> constant curvature<br />
<strong>space</strong>s. Trans. Amer. Math. Soc., 358(3):1105–1115 (electronic), 2006.<br />
F.W. Warner. Foundations of differentiable manifolds and Lie groups. Scott, Foresman<br />
and Co., Glenview, Ill.-London, 1971.
Notation<br />
( , ) : Hermitian product <strong>in</strong> C n+1 def<strong>in</strong>ed at (1.2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9<br />
( , ) ɛ : Hermitian product <strong>in</strong> CK n (ɛ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10<br />
〈 , 〉 ɛ : Hermitian metric of CK n (ɛ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10<br />
α i : real part of a dual form of a J-mov<strong>in</strong>g frame on CK n (ɛ) . . . . . . . . . . . . . . . . . . . . . . . . . 16<br />
α ij : real part of a connection form of a J-mov<strong>in</strong>g frame on CK n (ɛ) . . . . . . . . . . . . . . . . . . 16<br />
B k,q (Ω) : Hermitian <strong>in</strong>tr<strong>in</strong>sic volume . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38<br />
β i : imag<strong>in</strong>ary part of a dual form of a J-mov<strong>in</strong>g frame on CK n (ɛ) . . . . . . . . . . . . . . . . . . . 16<br />
β ij : imag<strong>in</strong>ary part of a connection form of a J-mov<strong>in</strong>g frame on CK n (ɛ) . . . . . . . . . . . . 16<br />
C n : standard Hermitian <strong>space</strong> . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8<br />
CH n : <strong>complex</strong> hyperbolic <strong>space</strong> . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8<br />
CK n (ɛ) : <strong>complex</strong> <strong>space</strong> form with constant holomorphic curvature 4ɛ . . . . . . . . . . . . . . . . . 8<br />
CP n : <strong>complex</strong> projective <strong>space</strong> . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8<br />
cos ɛ (α) : generalized cos<strong>in</strong>e function with angle α . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10<br />
cot ɛ (α) : generalized cotangent function with angle α . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10<br />
D :<br />
distribution def<strong>in</strong>ed by the normal vector and the <strong>complex</strong> structure <strong>in</strong> a hypersurface<br />
<strong>in</strong> a Kähler manifold . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40<br />
dL r : density of the <strong>space</strong> of <strong>complex</strong> planes with <strong>complex</strong> dimension r . . . . . . . . . . . . . . . 20<br />
E : bisector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82<br />
Γ k,q (Ω) : Hermitian <strong>in</strong>tr<strong>in</strong>sic volume . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38<br />
G C n,r : Grassmannian of <strong>complex</strong> r-planes <strong>in</strong> C n . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47<br />
G n,k,p (C n ) : Grassmannian of (k, p)-planes <strong>in</strong> C n . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81<br />
H n : real hyperbolic <strong>space</strong> . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8<br />
H : sub<strong>space</strong> of C n+1 which def<strong>in</strong>es the po<strong>in</strong>ts <strong>in</strong> CK n (ɛ) <strong>in</strong> the projective model . . . . . . 9<br />
J : <strong>complex</strong> structure of a <strong>complex</strong> manifold . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7<br />
K(V ) : convex compact doma<strong>in</strong>s <strong>in</strong> the vector <strong>space</strong> V . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29<br />
107
108 Notation<br />
L R r : <strong>space</strong> of totally real planes with dimension r <strong>in</strong> CK n (ɛ) . . . . . . . . . . . . . . . . . . . . . . . . . . 19<br />
L R r : totally real plane <strong>in</strong> L R r . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19<br />
L C r : <strong>space</strong> of <strong>complex</strong> planes with <strong>complex</strong> dimension r <strong>in</strong> CK n (ɛ) . . . . . . . . . . . . . . . . . . . 19<br />
L C r : <strong>complex</strong> plane <strong>in</strong> L C r . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19<br />
L (r)<br />
q : <strong>space</strong> of <strong>complex</strong> q-planes conta<strong>in</strong>ed <strong>in</strong> a fixed <strong>complex</strong> r-plane . . . . . . . . . . . . . . . . . 25<br />
: <strong>space</strong> of <strong>complex</strong> r-planes conta<strong>in</strong><strong>in</strong>g a fixed <strong>complex</strong> q-plane . . . . . . . . . . . . . . . . . . 25<br />
L C r[q]<br />
M n×n (C) : square matrices of order n with <strong>complex</strong> valued entries . . . . . . . . . . . . . . . . . . . 11<br />
M i (S) : i-th mean curvature <strong>in</strong>tegral of a hypersurface S . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30<br />
M D i<br />
(S) : i-th mean curvature <strong>in</strong>tegral restricted to the distribution D . . . . . . . . . . . . . . . . 40<br />
µ k,q : Hermitian <strong>in</strong>tr<strong>in</strong>sic volumes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39<br />
N(Ω) : normal fiber bundle of a doma<strong>in</strong> Ω ⊂ CK n (ɛ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16<br />
O n : area of the sphere of radius 1 <strong>in</strong> the standard Euclidean <strong>space</strong> of dimension n . . . 26<br />
ϕ i : dual form of a J-mov<strong>in</strong>g frame on CK n (ɛ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16<br />
ϕ ij : connection form of a J-mov<strong>in</strong>g frame on CK n (ɛ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16<br />
P U(n) : projective unitary group . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11<br />
RH n : real projective <strong>space</strong> . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8<br />
S(CK n (ɛ)) : unit tangent bundle of CK n (ɛ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15<br />
σ i (A) : i-th symmetric elementary function of the bil<strong>in</strong>eal form A . . . . . . . . . . . . . . . . . . . . 30<br />
s<strong>in</strong> ɛ (α) : generalized s<strong>in</strong>e function with angle α . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10<br />
U k,p : valuation def<strong>in</strong>ed on C n . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34<br />
U(p, q) : unitary group of <strong>in</strong>dex p, q . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11<br />
U ɛ (n) : isometry group of CK n (ɛ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11<br />
V i (Ω) : i-th <strong>in</strong>tr<strong>in</strong>sic volume of the doma<strong>in</strong> Ω ⊂ R n . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29<br />
Val(V ) : cont<strong>in</strong>uous translation <strong>in</strong>variant valuations on the vector <strong>space</strong> V . . . . . . . . . . . 31<br />
Val k (V ) : homogeneous valuations of degree k <strong>in</strong> Val(V ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31<br />
Val + (V ) : even valuations <strong>in</strong> Val(V ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31<br />
Val − (V ) : odd valuations <strong>in</strong> Val(V ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31<br />
Val G (V ) : <strong>in</strong>variant valuations under the action of the group G <strong>in</strong> Val(V ) . . . . . . . . . . . . 31<br />
Val U(n) (C n ) : cont<strong>in</strong>uous translation and U(n)-<strong>in</strong>variant valuations on C n . . . . . . . . . . . . 34<br />
ω n : volume of the ball of radius 1 <strong>in</strong> the standard Euclidean <strong>space</strong> of dimension n . . . 38<br />
ω i0 : dual form of an orthonormal mov<strong>in</strong>g frame on CK n (ɛ) . . . . . . . . . . . . . . . . . . . . . . . . . . 16<br />
ω ij : connection form of an orthonormal mov<strong>in</strong>g frame on CK n (ɛ) . . . . . . . . . . . . . . . . . . . . 16
Index<br />
2-po<strong>in</strong>t homogeneous <strong>space</strong>, 14<br />
3-po<strong>in</strong>t homogeneous <strong>space</strong>, 14<br />
aff<strong>in</strong>e surface area, 30<br />
Alesker, 29, 32, 34, 36, 37, 58<br />
almost <strong>complex</strong> structure, 7<br />
angle between two planes, 9<br />
Bergmann metric, 10<br />
Bernig, 36–38, 61<br />
bisector, 18, 82, 83<br />
Blaschke, 29<br />
cohomogenity 1 hypersurface, 83<br />
coisotropic plane, 88<br />
<strong>complex</strong><br />
hyperbolic <strong>space</strong>, 8<br />
manifold, 7<br />
projective <strong>space</strong>, 7, 8<br />
<strong>space</strong> form, 7, 8, 14, 18, 36<br />
sp<strong>in</strong>e of a bisector, 82<br />
structure, 7<br />
submanifold, 17<br />
sub<strong>space</strong>, 17, 81<br />
cont<strong>in</strong>uous valuation, 30<br />
Crofton formula, 32<br />
even valuation, 31<br />
Fu, 36–38, 61<br />
Fub<strong>in</strong>i-Study metric, 10<br />
generic position, 48<br />
geodesic ball, 18<br />
Hadwiger, 32<br />
Hadwiger Theorem, 32, 36, 37<br />
Hausdorff<br />
distance, 30<br />
metric, 30<br />
Hermitian<br />
<strong>in</strong>tr<strong>in</strong>sic volume, 40, 61<br />
product, 7<br />
holomorphic<br />
angle, 9<br />
atlas, 7<br />
curvature, 8<br />
section, 8<br />
homogeneous<br />
hypersurface, 83<br />
<strong>space</strong>, 13, 19<br />
valuation of degree k, 31<br />
horizontal lift, 10<br />
<strong>in</strong>tr<strong>in</strong>sic volume, 29–31<br />
<strong>in</strong>variant<br />
density, 19, 20<br />
with respect to a group, 31<br />
isometry group, 11, 14, 34<br />
isotropy group of a <strong>complex</strong> r-plane, 19<br />
J-bases, 14<br />
J-mov<strong>in</strong>g frame, 14<br />
(k, p)-plane, 81<br />
Kähler<br />
form, 8<br />
manifold, 8<br />
k<strong>in</strong>ematic formula, 33<br />
Lagrangian plane, 36, 90<br />
l<strong>in</strong>ear<br />
(k, p)-plane, 81<br />
submanifold, 81<br />
McMullen, 32<br />
Theorem, 31<br />
mean curvature <strong>in</strong>tegrals, 30, 31, 33<br />
monotone valuation, 31<br />
mov<strong>in</strong>g frame, 14<br />
odd valuation, 31<br />
Park, 38–40, 75<br />
Quermas<strong>in</strong>tegrale, 26<br />
109
110 Index<br />
real<br />
hyperbolic <strong>space</strong>, 8<br />
projective <strong>space</strong>, 8<br />
<strong>space</strong> form, 8, 18<br />
real sp<strong>in</strong>e of a bisector, 82<br />
regular doma<strong>in</strong>, 16<br />
reproductive<br />
formula, 33<br />
property, 45, 56<br />
restricted mean curvature <strong>in</strong>tegral, 40<br />
r-pla <strong>complex</strong>, 17<br />
r-pla totalment real, 17<br />
Rum<strong>in</strong><br />
derivative, 61<br />
operator, 61<br />
second fundamental form, 30<br />
slide of a bisector, 82<br />
smooth valuation, 31<br />
on a manifold, 37<br />
<strong>space</strong> constant<br />
holomorphic curvature, 7, 8, 14, 18<br />
sectional curvature, 8, 18<br />
sp<strong>in</strong>al surface, 82<br />
standard Hermitian <strong>space</strong>, 8<br />
Ste<strong>in</strong>er formula, 29, 30<br />
symmetric elementary function, 30, 40<br />
symplectic form, 38<br />
template method, 33<br />
totally<br />
geodesic<br />
submanifold, 17<br />
sub<strong>space</strong>, 81<br />
real<br />
plane, 9<br />
submanifold, 17<br />
sub<strong>space</strong>, 17, 81<br />
umbilical hypersurface, 18<br />
translation <strong>in</strong>variant valuation, 31<br />
trigonometric generalized functions, 10<br />
unimodular group, 13<br />
unit normal bundle, 16, 17<br />
unit tangent bundle, 15<br />
U(p, q), 11<br />
valuation, 29